Pharmacokinetic studies in cancer chemotherapy: Usefulness in clinical practice

Pharmacokinetic studies in cancer chemotherapy: Usefulness in clinical practice

Cancer Treatment Reviews (1997) 23, 153-169 Pharmacokinetic studies in cancer chemotherapy: usefulness in clinical practice G. Freyer*, B. Lignea...

1MB Sizes 1 Downloads 72 Views

Cancer

Treatment

Reviews

(1997)

23, 153-169

Pharmacokinetic studies in cancer chemotherapy: usefulness in clinical practice G. Freyer*, B. Ligneau*, and V. Trillet-Lenoir*

B. Tranchandt,

C. Ardiett,

F. Serre-Debeauvaiss

*Medical Oncology Unit, Radiotherapy and Oncology Department, Hospitalier Lyon-Sud, France t Pharmacokinetics Unit, Centre LBon BBrard, Lyon, France 4 Pharmacokinetics Unit, Hopital Michallon, Grenoble, France

Centre

Introduction During the last two decades, survival in most cancers has not been significantly improved by chemotherapy. Although major survival gains have been obtained in haematopoietic malignancies or germ cell tumours, the prognosis of the most frequently diagnosed tumours (colon-rectum, breast, lung) remains poor for patients with advanced diseases. Anticancer chemotherapy efficiency might be improved in the future by new cytotoxic agents such as taxoids, topoisomerase-1 inhibitors, new alkaloids and antimetabolites, but another promising research field emerges from the better knowledge of systemic exposure to cytotoxic drugs. Pharmacokinetics is the study of metabolism of drugs in biological fluids, tissues and excreta (I). For a given dose, absorption, distribution, metabolism and excretion of a drug varies from one patient to another, depending on many physiological and pathological factors. Significant clinical variability is observed among patients treated by chemotherapy, both in terms of toxicity and efficacy (2). Such inter-patient variations of the pharmacokinetic features might determine the known clinical variability among patients treated by chemotherapy. The purpose of this paper is to review the data on the ability of pharmacokinetic studies to determine the causes and levels of clinical variability and how they might improve the clinical benefit from anticancer treatments. Variability Main pharmacokinetic

of pharmacokinetic

variables

(PK) parameters

(3)

The usual route of anticancer chemotherapy administration (either bolus or continuous infusion), the use of the oral 0305-7372/97/030153

+ 17 $12.00/O

0 1997 153

W.B.

is intravenous route being less Saunders

Company

Ltd

154

G. FREYER

HAL.

Y L

I

Time Figure 7. Rapid intravenous concentration rapidly reaches

Time injection in a three-compartment its maximal value and decreases concentration.

model. in three

After i.v. injection, plasma phases (a, B, y). Cp, plasma

common. In Figure 1, a cytotoxic agent has been administered as a rapid intravenous injection. The drug concentration is assayed at defined intervals of time in the plasma, and the values are then extrapolated using mathematical models. For the majority of anticancer drugs, plasma concentrations first reach a maximal value (Cmax), then decrease in one, two or three phases (multicompartimental models). Each of these phases is characterized by a specific half-life. Useful PK parameters are described below. The half-life of a drug is defined as the period of time after its administration for the concentration to be decreased by half. It may be increased in the case of renal or hepatic dysfunction. Clearance (CL) and area under curve (AUC) are the main pharmacokinetic parameters for clinical purposes, since they are strongly correlated to drug exposure in a given patient. Clearance is the apparent plasma volume cleared of the drug per unit of time. It takes into account various means of elimination (hepatic, renal, pulmonary), and describes the patient’s capability to eliminate the drug. As shown in Figure 1, the AUC is a function of both plasma concentration and time (from the beginning of the injection or infusion to infinity); the AUC increases when drug elimination (or clearance) decreases, according to the following formula: CL (ml/min)

= Dose (mg)/AUC

(mg/ml

min)

Based on the theoretical assumption that the drug’s distribution in a given volume is homogeneous, its concentration in this volume is equivalent to the overall plasma concentration, but the so-called ‘apparent volume of distribution’ (Vd) is neither a real nor a physiological entity. In fact, the drug’s distribution in the various body compartments depends on many different factors which are related to the drug’s properties, such as protein binding or liposolubility. The Vd can be expressed in either l/m* or l/kg. For instance, anthracyclines

PHARMACOKINETIC

STUDIES

IN CANCER

have a rather large Vd due to their high affinity example, to melphalan (4, 5).

Parameters

of pharmacokinetic

Pharmacokinetic variability pathological characteristics, administration. Patients’

physiological

CHEMOTHERAPY

for fat tissues,

variability

155

in contrast,

for

in cancer patients

may be related to patients’ physiological and/or and/or to the drug dosages and methods of

conditions

Although chronological age is the most significant indicator of aging, some studies have suggested the relevance of ‘biological age’ (6). The World Health Organization’s definition of ‘old’ for a human being is a chronological age over 70 years. This definition does not take into account inter-individual physiological variability and should not be used as the sole criterion for therapeutic decisions. Pharmacokinetic variability appears to be higher in the elderly. Furthermore, glomerular filtration impairment and/or hypoprotinaemia frequently occur over 60 years of age; these modifications probably account, at least partly, for this variability (7). Elderly patients are frequently offered sub-optimal cancer treatments, even for potentially curable diseases (8, 9). According to the NCI-EORTIC (National Cancer Institute-European Organization for Research and Treatment of Cancer) consensus, elderly patients should not be excluded from cancer research treatment programmes and should as often as possible participate in clinical trials which include pharmacokinetic studies (IO). In patients under the age of 15 years, variations in systemic exposure and clearances of cytotoxic drugs have been shown to be functions of age, in particular for methotrexate, ifosfamide and carboplatin (1 l-13). Variations in body composition and especially weight loss may result in modifications of Vd for lipophilic drugs such as anthracyclines, and in variations of the plasma exposure as reflected by the AUC. Major pharmacokinetic variability has also been observed for non-lipophilic drugs. For instance, the half-life of ifosfamide is increased in obese patients, suggesting alterations of the liver function by currently unknown mechanisms (14). Pathological

conditions

Renal dysfunction. Renal excretion is an important route of elimination for several drugs (methotrexate, cisplatin, carboplatin, cyclophosphamide, bleomycin, hydroxyurea, etoposide, streptozotocin). Any alteration of renal function may induce drug accumulation and increased toxicity, indicating a need for dose adjustment. Dose reduction is usually recommended for drugs which themselves or their active metabolites are excreted for more than onethird by the kidney when the glomerular filtration is less than 70% of normal value (15, 16). Among the various drugs routinely used, cisplatin is one of the most nephrotoxic. In order to minimize this limiting toxicity, the development of cisplatin analogues was initiated in the early 1980s and led to the synthesis

156

G. FREYER

ETAL.

of carboplatin. This drug is eliminated by glomerular filtration only and is less nephrotoxic than cisplatin. Its equivalence to cisplatin in terms of efficacy has been demonstrated in tumour types such as ovarian cancer (17). Models of dose-adaptation models to renal function have been described for methotrexate and bleomycin (18, 19). The various effects of liver dysfunction on plasma drugs Liver dysfunction. levels are not easily predictable by functional or biochemical tests. In the presence of a hepatocellular defect, the use of drugs which are eliminated by intra-hepatic biotransformation (such as oxazaphosphorines or anthracyclines) may require dose adjustment. Some studies have demonstrated the usefulness of dose adjustment with abnormal hepatic function tests for doxorubicin (21, teniposide (20) and etoposide (21). Biliary obstruction induces the accumulation of drugs which are excreted unchanged by this route, such as vinca alkaloids (vincristine, vinblastine, vindesine, vinorelbine) (22). This may result in increased haematological and neurological toxicity, and justify dose reduction. The cytotoxic activity and/or elimination of many cytotoxics require various intra-hepatic enzymatic bio-transformations: taxoids are metabolized by cytochrome ~450; 5-fluorouracil is converted into fluorodihydrouracil by dihydropyrimidine dehydrogenase, an enzyme with marked circadian variations. Due to constitutional differences in enzymatic expression, there is a high inter-individual heterogeneity in the ability to activate and/or inactivate cytotoxic drugs (23). Hypoprotinaemia. Some drugs are strongly bound to plasma proteins. Since only the unbound fraction (the so-called ‘free-drug’) is able to diffuse from the plasma compartment to the tumour interstitium, the cellular effects of a drug are more closely correlated to the free-drug level than to the total drug concentration. Hypoprotinaemia may occur in various clinical situations such as liver diseases, renal failure and malnutrition, and result in unexpected chemotherapy-induced toxicity. This phenomenon is restricted to drugs showing a high affinity for proteins, such as cisplatinum. Presence of a third compartment. The occurrence of ascitis, oedema or pleural effusion may delay the elimination of cytotoxic drugs, such as methotrexate. and therefore increase their toxicity (24). Variations of hoavailability. When a drug is given orally, its plasmatic level depends on its absorption by the gastrointestinal tract and may be modified by pH variations, simultaneous intake of other drugs and alterations of visceral motility. For instance, any activation of intestinal motility results in increased clearance of methotrexate; conversely, the paralytic ileus induced by vincristine may reduce the clearance of concomitant methotrexate. Cyclophosphamide and etoposide may be prescribed by oral route, but their plasma concentrations may show marked variations, with a risk of decreased efficacy in some of the patients treated.

PHARMACOKINETIC

Drug dosage

STUDIES

IN CANCER

CHEMOTHERAPY

157

and administration

Body-surface area and dose adaptation. It is routine practice in medical oncology to determine chemotherapy doses on the basis of calculated bodysurface area (BSA) rather than weight. This method was suggested by Pinkel in the 1950s because of the known correlation between BSA and blood volume in rats (25). Its relevance has never been extensively studied in cancer patients, and some authors suggest that it could be responsible for unexpected variations in plasma concentrations of cytotoxic drugs. Moreover, the DuBois formula for the calculation of BSA was established in 1916 on the basis of post-mortem body measurements performed on eight adults and one child, and is of doubtful validity (26). Last but not least, BSA does not take body composition into account; this is a key parameter for the distribution profile of several drugs (such as anthracyclines in fat tissue). High-dose chemotherapy. Dose escalation of cytotoxic drugs may induce changes in pharmacokinetic parameters and lead to non-linear elimination. If so, the parameters that have been determined for ‘standard doses’ cannot be used for high-dose chemotherapy. This has been shown for 5-fluorouracil (5FU) and ifosfamide but not for melphalan, for which the test-dose method (vi) may be used (5). Drug interactions Concomitant use of various cytotoxic drugs in polychemotherapy regimens increases the risk of unexpected interactions as well as associated treatments such as anti-emetics, antibiotics, anxiolytics and analgesics. Balis et al. have reviewed some of these interactions, but many of them are probably still to be discovered (27).

Is pharmacodynamic

(PD) variability

related to PK variability?

The clinical variability induced by cytotoxic treatments is well known by physicians, both in terms of efficacy and toxicity. Pharmacodynamics involves the study of drug effects on tumour cells as well as normal cells, and their clinical consequences. From a PD point of view, the variability of the cytotoxic effects may be due to various factors such as tumour vascularization abnormalities, pleiotropic resistance to cytotoxic agents, active cellular processes of elimination or tumour heterogeneity. Those mechanisms are still poorly understood and are not the purpose of this review (28). On the other hand, PK variability may be quantified and may sometimes at least partly explain the high degree of clinical heterogeneity from one patient to another. A large amount of data are available on PK/PD relationships, and evidence of PK/PD correlations has been demonstrated for various cytotoxic agents. Pharmacokinetic studies designed to look for PWPD correlations require precise technical conditions and careful selection of the different parameters. The most relevant compound to be assayed may be either the administered

158

G. FREYER

ETAL.

drug itself or one of its active metabolites, as demonstrated by Rustum et a/. (29) for cytarabine (Ara-C) and its active metabolite, Ara-CTP. As the assay must be easily reproducible in routine practice, the most frequently tested biological fluid for PK studies is plasma. However, in leukaemia patients treated with AraC, the intracellular concentrations of Ara-CTP in the blast cells are more strongly correlated with the therapeutic effect than the plasma concentrations. The doses of cytotoxic drugs must be precisely calculated and totally administered to the patient by regular flow infusion. The use of electronic pumps is recommended. The precise timing of each blood sample must be respected. The blood sampling procedures should be compatible with both the patient’s quality of life and the nurses’ work schedule. Extensive knowledge of the PK properties of each new drug should be provided by PK studies performed during Phase I-II development, in order to determine the optimal systemic exposure (defined by maximal tumour response and minimal toxicity). National Cancer Institute and EORTC recommendations suggest the use of PK during dose escalation in Phase I in order to reach the maximal tolerated systemic exposure (MTSE) rather than the maximal tolerated dose (MTD) (30). Early PK data could then facilitate the PK approach for further studies. As the assessment of tumour response and cumulative toxicity requires more than one chemotherapy course, PK studies should be performed with each course for a given patient. Various methodological difficulties may be encountered in the demonstration of a correlation between the. PK indicators of the plasma exposure to the anticancer agents and the PD endpoints. The mean number of chemotherapy courses in a patient is around six, and various drugs are generally injected to the patient according to different schedules. Should the PK parameters be determined for each drug at each course or could it be extrapolated from the results of the first or two first courses if the intra-patient variability is neglectable? Should the tumour response be regarded as a continuous variable expressed as a percentage of tumour regression or more often as discontinuous [according to World Health Organization (WHO) criteria] and divided into complete response, partial response, stable disease and progressive disease? The same problem arises for the evaluation of the toxicities: some parameters such as blood counts can be expressed on a continuous scale whereas others are necessarily discontinuous (such as WHO toxicity grades for clinical secondary effects). The high number and heterogeneity of these parameters is responsible for the complexity of such statistical analyses and for the limitation of PK/ PD correlations. However, some useful mathematical models are available. Discontinuously scaled parameters of patients with or without toxic adverse effects can be compared using statistical tests such as the Spearman rank test or Student’s t-test, but the usefulness of this approach is limited since most patients experience toxicity. Linear regression analysis and computerization of the correlation coefficient between PK and PD parameters may also be performed. Despite apparent simplicity, this method is not well adapted to the description of PK/PD relationship in oncology, since it measures the degrees of association between the parameters rather than indicates how well estimations match true values (31). The sigmoidal maximal effect (Emax) describes the PK/

PHARMACOKINETIC

STUDIES

IN CANCER

CHEMOTHERAPY

159

is 5 50% Emax --~---_----_--

Pharmacokinetic Figure 2. The sigmoidal endpoint (such as tumour

Table

1.

Cytotoxic

Cisplatinum Carboplatin Methotrexate 5-FU Adriamycin Epirubicin Etoposide Vincristine Paclitaxel

Relationship anticancer

agent

parameter

(P)

maximal effect model (Emax). In this example, E could response rate in a population of patients) and P could exposure of a cytotoxic agent (area under curve).

between drugs

pharmacokinetic

parameters,

efficacy

and

toxicity

be the dynamic be the plasma

for some

major

Efficacy (PK parameters)

Toxicity (PK parameters)

Stat.

Refs

Cp, AUC AUC CP Cp, AUC

NV, R: Cp, AUC AUC Neu, Thr NV, Neu, HFS: Cp, AUC, Cl Neu, Card: Cp, Css Neu: AUC Neu, Thr: Cp, AUC NI Neu: AUC

LinReg. Student-Hill Hill Student Lin.Reg. Spearman Lin.Reg. Lin.Reg. Lin.Reg. M.W.U.T. Hill

67 67 17, 69 48, 55, JO 56,Jl 57 72, 73

CP,Cl AUC Cp, AUC -

Cp, plasma concentration; AUC, area under curve; NV, nausea and vomiting; Neu, neutropenia; Thr, thrombocytopenia; NI, neurologic toxicity; HFS, hand-foot clearance; Card, cardiotoxicity; Css, steady-state concentration; Stat, statistical linear regression; Student, Student’s t-test; Hill, fitted to Hill’s equation; Spearman, test; M.W.U.T., Mann-Whitney U-test.

PD relationship on the basis of in vitro models situations, according to the Hill equation:

extrapolated

1

JO, 71 22 42 R, renal toxicity; syndrome; Cl, model; Lin.Reg., Spearman rank

to the clinical

E =Emax.(P)H/(P50)H + (PJH where E is a continuously scaled PD effect, Emax is the maximum elicitable effect, P is a pharmacokinetic parameter (Cmax or AUC), and P, is the value of P which results in 50% of the Emax (Figure 2). H is the Hill’s constant which defines the degree of sigmoidity of the model. This model is well adapted to oncological clinical research practice (32, 33). Several authors have shown strong correlations between PK parameters of anticancer drugs and PD results. Table 1 gives the most significant parameters for some of the major cytotoxic drugs and the statistical methods used for

160

G. FREYER

ETAL.

each PK/PD correlation. Since toxicity is the result of cytotoxic effects on normal cells and tissues, it may be more clearly correlated to the PK parameters, in contrast to the tumour response which depends on cell-related mechanisms of resistance to the drugs. Indeed, for the same plasma exposure, the PD results may be significantly increased in the normal cells compared to the tumour cells. This difference may explain the greater number of PK/PD correiations based on toxicity published to date of cytotoxic agents compared to the data on their efficacy. As shown in Table 1, AUC is the most relevant parameter in various situations. For some specific anticancer agents which act on a specific phase of the cell cycle and are therefore administered as continuous infusions in order to maximize the probability of cytotoxicity, the time above a specific threshold concentration may be more critical. This has been well established for etoposide; the haematologic toxicity of this phase-specific agent is better correlated to its peak concentration than to the AUC, even if the latter parameter remains discriminant to predict the antitumour effect. In the case of carboplatin, the AUC is simply calculated when the glomerular filtration rate is known, because there is no tubular re-absorption or excretion for this drug. This property has enabled Calvert et al. to propose some formulae for use in clinical practice. PWPD correlations have been more extensively studied with recently available new drugs, and examples of docetaxel and irinotecan are given below. Dote taxel Docetaxel is an interesting new semi-synthetic taxoid with a high activity in many solid tumours. The main limit to its clinical use is its low therapeutic index. Most patients develop neutropenia and/or other complications including peripheral neurotoxicity and oedema. Human PKstudies were performed during Phase I and II trials. In those studies, docetaxel was administered as a short I- or 2-h infusion and plasma concentrations were measured using highperformance liquid chromatography (HPLC) (34, 35). Docetaxel has a tricompartimental pharmacokinetic profile at the usual therapeutic dosages, and the linearity of the pharmacokinetic parameters has been well established. Extra et a/. have shown a good correlation between the AUC of docetaxel and the nadir of neutrophils (34). In Phase II studies, Launay-lliadis et al. (36) and Bruno et al. (37) have found a correlation between the clearance of docetaxel and the risk of WHO Grade 4 neutropenia. Five prognostic co-variables have been identified to explain the high inter-individual variability of the clearance of docetaxel in 280 patients: -Age >70 years (6.7% decrease of the clearance). -Hypoalbuminaemia (10% decrease of the clearance). -Elevation of al-glycoprotein (>normal value+2.6g/l; 19% decrease of the clearance). -Elevation of alanine aminotransferase (ALAT) and/or aspattate aminotransferase (ASAT) and/or alkaline phosphatase (33% decrease of the clearance). -Body-surface area (BSA) (decrease of the clearance when BSA is low).

PHARMACOKINETIC Table

2.

STUDIES

IN CANCER

Mean SD cv (%)

Among these factors, discriminant (35, 37). lrinotecan

161

Inter-patient variability in SN-38 and CPT11 area under curve (AUC) in 47 patients AUC CPT-11 (mg/ml min)

Adapted

CHEMOTHERAPY

AUC SN-38 (mg/ml min)

25.243 7.475 29.6 from

496 444 89.5

ref. 42.

al-glycoprotein

and

liver

function

are

the

most

(CPT- II)

Irinotecan, a new camptothecin derivative, has demonstrated antitumour activity in various tumour types, mediated by a topoisomerase 1 inhibition. In vivo, CPT-11 is converted into SN-38 (7-ethyl-IO-hydroxy camptothecin), the main active metabolite which is at least lOO-fold more active than CPT-11 itself (38). Pharmacokinetic studies conducted Phase I trials indicated a linear relationship between the AUC of SN-38 and the administered dose of CPT-11, whatever the therapeutic schedule (once daily for 3 consecutive days every 3 weeks, or once weekly for 3 consecutive weeks, or once every 3 weeks) (39-41). A PKstudy was subsequently performed by Canal eta/. (42) during a Phase II trial, to investigate inter-patient and intra-patient variability and PWPD relationships. They studied 47 patients with metastatic colorectal cancer treated by a 30 min intravenous infusion of CPT-11 350 mg/m’ every 3 weeks. CPT-11 and its two metabolites, namely SN-38 and SN-38 glucuronide (SN 28 G), were determined by HPLC. A high inter-individual metabolic variability was found by the authors since the coefficient of variation of the ratio (AUC SN-38+AUC SN-38 G)/AUC CPT-11 was 51.6%. The same level of variability was shown in the plasma exposure to SN-38 (Table 2). In the same study, a significant relationship was found between the percentage of reduction of the neutrophil count and the AUC of CPT-11 (t-=0.597, pcO.001) and SN-38 (r=0.559, pcO.001). On the other hand, no relationship was found between any PK parameter and the tumour response rate (42).

Potential Dose adaptation

clinical interest

in oncological

of PK dose adaptation

practice

Individual dose adaptation of anticancer drugs is of potential interest in terms of efficacy and toxicity. As already mentioned, dosages are usually established according to BSA, without taking into account the above-reviewed biological and physiological inter-patient variability parameters. Consequently, the AUC of a given dose/m* of a drug may vary up to IQ-fold from one patient to another (43). Dose adaptations are usually determined according to the level of toxicity

162

G. FREYER

ET AL.

observed during the previous course(s). In a population of patients, different subgroups experiencing either little or severe toxicity may be identified. In the first one, the maintenance of a constant dose level while it could be safely increased might jeopardize an optimal antitumour effect. In the second, unacceptable toxicity may lead to empirical dose reductions which might also compromise the therapeutic result (44). In retrospective studies, Hryniuk et a/. (45, 46) have shown that any decrease in the dose intensity of a chemotherapy regimen as compared to the reference protocol may reduce the clinical benefit. The determination of target PK parameters such as AUC and consequent dose adjustment for each individual might reduce the clinical variability and increase the therapeutic index. For example, in chemosensitive and potentially curable diseases, the validity of the concept of dose intensity remains questionable. Furthermore, it could very well be that the lack of correlation between chemotherapy dose intensification and patients’ survival in most tumour models could be due to high inter-individual PK variability rather than to the irrelevance of the dose-effect relationship concept. Indeed, a 50% increase in the overall dose intensity of a given chemotherapy protocol used in a population of patients may result, for a given patient, in a substantial and unintended under-treatment due to low plasma exposure to the drugs that could have been predicted and may be avoided by a PK approach. As suggested by Evans et a/. (281, AUC intensity (defined as the ratio AUC/time between two consecutive courses) could be a more relevant parameter of the level of intensity of chemotherapy regimens than administered dose intensity. Hence, recently developed PK doseadaptation strategies are of high potential interest in oncology. PK dose-adaptation

strategies

Despite the major above-listed methodological difficulties, some methods are now available that allow pharmacokinetically-guided cytotoxic treatments. The most well-known example is the measurement of methotrexate residual concentrations after intravenous injection, in order to avoid myelotoxicity by the use of folinic acid rescue (47, 48). Individual

dose adaptation

based on clinical and biological

characteristics

The main interest of this method is to allow the calculation of the chemotherapy dose without the requirement of any additional blood sample. In this case, PK parameters and their relationship to individual morphological and biological characteristics must have been established previously. The most important example in this area is the carboplatin dose-adaptation method forthe treatment of ovarian cancer. The AUC of total carboplatin is directly related to its efficacy and toxicity, as shown by Jodrell et a/. (17). One should keep in mind that only the free-carboplatin fraction should be considered as active, and that the freecarboplatin AUC is closely related to total carboplatin AUC in all the available studies. Furthermore, carboplatin clearance is linked to the glomerular filtration rate (GFR) but not to tubular re-absorption or excretion. Calvert et al. showed that radionucleide EDTA clearance allows a reliable estimation of GFR (15). Optimal AUC, defined as AUC associated with the best tumour response/toxicity

PHARMACOKINETIC

STUDIES

IN CANCER

CHEMOTHERAPY

163

ratio, were determined on the basis of clinical studies using carboplatin either as a single agent or combined with other drugs. The objective of the study performed by Calvert et a/. (7) was to determine a calculation method of carboplatin doses allowing optimal exposure as defined by optimal AUC (or ‘target AUC’). They described the following formula: Dose=target

AUC x (GFR+25)

This formula suggests that higher doses may be required to reach optimal AUC for patients with higher GFR. In order to obtain a manageable degree of myelosuppression, the proposed AUC is 5 mg/mI min for previously treated and 7 mg/mI min for previously untreated patients (17). However, the determination of EDTAclearance is not of routine use. Therefore, there has been a tendency among clinicians to use the Calvert formula based on the Cockcroft method for the calculation of creatinine clearance: CrCl=1.23x(140-A)xWt/SCr CrCl = 1.04 x (140 -A)

in men x Wt/Scr

in women

where CrCl is creatinine clearance (ml/min), A is age (years), Wt is weight (kg), and Scr is serum creatinine (mg/l). The lack of adequate correlation between this method and the radionuclide EDTA test for the calculation of GFR may induce false results and therefore lead to either unexpected toxicities or lack of efficiency (15). To avoid this problem, Chatelut et a/. (49) have recently described a new formula based on a population PK model (non-linear mixed effect model, NONMEM). This formula allows the calculation of carboplatin doses for a given patient using serum creatinine, age, sex, weight and target AUC (49). This method is much easier to apply in clinical routine use and the predicted values of carboplatin clearance are not statistically different from those obtained by the radionuclide EDTA method: Cl carboplatin

=0.134 x Wt+[(218 x Wt) x (I -0.00457 Dose (mg) =desired AUC x Cl carboplatin

x A) x Cr x Sl

where S is a correction coefficient (S = 1 if male; S =0.686 if female). Using the Calver-t formula (with radionuclide EDTA estimation of the GFR) or the Chatelut formula, the clinician is able to reach the optimal plasma exposure to carboplatin for most patients. However, for some categories of patients, such as the obese (weight >I30 kg), the Chatelut formula tends to be less precise and should therefore be used with caution. On the other hand, it allows a good estimation of carboplatin clearance in case of renal failure. Recently, Calvert et al. suggested a new limited-sampling strategy which also appears to be promising (30). Dose adaptation

by limited

sampling

Residual concentration (Cmin) may be used in order to adapt the doses of 6mercaptopurine (51). Cmax is also relevant for methotrexate, etoposide and

164

G. FREYER

ET AL.

Time Figure

3.

Continuous

infusion

and

steady-state

concentration

(Css).

Cp,

plasma

concentration.

teniposide (1, 52, 53). The concentrations at 24 and 48 h have been used to estimate the risk of methotrexate toxicity by Staller et a/. and Evans et al. (54, 55). The calculation of the AUC requires numerous blood samples at each cycle, in order to adapt the dose of cytotoxic drug from one course to another on the basis of previously established correlations with toxicity and/or efficiency. It has been successfully used for vincristine, 6-mercaptopurine, 5-fluorouracil and teniposide (1, 56). During continuous intravenous infusion (Figure 3), PK parameters may be determined from repeated plasma samples at each chemotherapy course. For the first course, chemotherapy is administered at a standard dose. The dose is then adapted to reach optimal systemic exposure (57). As described above, such an approach requires prior determination of optimal PK parameters. In a retrospective study, Gamelin et al. showed that the so-called FUFol regimen consisting of a weekly 8-h continuous infusion of 5-FU (1300mg/m’? in combination with folinic acid produces the best response rates with acceptable toxicity when the steady-state concentration (Css) is between 2 and 3 mg/l. They subsequently conducted a Phase II study in metastatic colorectal cancer using individual dose adaptation for 5-FU only in a weekly FUFol regimen as follows. The first course of 5-FU was given at the dose of 1300 mg/m’ and the dose was adapted weekly according to the PK parameters determined by limited sampling strategy. This dose-adaptation model allows a significant increase in the dose intensity of 5-fluorouracil as compared with the standard weekly FUFol regimen, as well as with the regimen described by Machover (5-FU: 450 mg/ m2; folinic acid: 200 mg/m’ Days l-5 every 4 weeks). In this study, the objective response rate was 54% and the toxicity of the FUFol regimen was as expected (58). These promising results raise the hypothesis of a possible dose-effect relationship for 5-FU in colorectal cancer (57). A Phase III randomized study is being conducted in order to compare this dose-adaptation strategy to the standard weekly regimen with fixed doses of 5-FU. Test-dose method This is an original way to determine individual PK parameters using multiple sampling after the administration of a very low dose of a cytotoxic drug. Optimal

PHARMACOKINETIC

STUDIES

IN CANCER

CHEMOTHERAPY

165

dose is then calculated according to PK parameters. Successful applications of the test-dose method have been published for methotrexate (59, 60) and melphalan (51, and intensive chemotherapy regimens could be used with reduced toxicity. To use this method, the linearity and stability of the PK parameters are required. Indeed, the parameters determined with the test dose of cytotoxic drug must not be modified when the higher dose is administered to the patient. The required number of blood samples is high and the main limitation to the routine use of this method.

Population

pharmacokinetic

models

(PPKM)

A PPKM is based on the previous description of inter-individual PK variability in a given population of patients (61). It therefore requires a preliminary step of individual sampling in order to estimate and quantify this variability, then to identify correlations between the PK profiles of the cytotoxic drugs and the patient’s specific co-variables (age, weight, creatinine, concomitant medications). Once the data from the overall population have been collected, the PK parameters of each new patient may be calculated with one to three concentration points. The main interest of this approach is to reduce the number of blood samples required to perform dose adaptation. Three types of method for PPKM may be distinguished. Data pool method. The principle of this method is to collect all the available data on the drug concentrations and to perform the statistical analysis as if those data had been obtained from a single patient. It is a simple approach but it does not allow a clear distinction between inter-individual and intraindividual variabilities. It is not commonly used. Two-step method. In the first step, PK parameters are calculated on an individual basis. In the second, population PK parameters are pooled in order to calculate the mean and the co-variance, and to set-up a mean-curve describing the population. The main disadvantage of this method is the overestimation of the variance of the PK parameters and consequent lack of precision in the calculation of the parameters for each new patient (47). Mixed-effect mode/s. These PPKM require a high level of technology from a mathematical point of view. Their application is facilitated by the use of specific software designed to estimate the patient’s variability as a density of probability function while taking into account individual co-variables. The use of this software requires specific training, but it is accessible to a motivated clinician. Among the most currently used, NONMEM (621, and NPEM (61) are adapted to oncological pharmacokinetic studies. Three cytotoxic drugs may be studied by PPKM in order to adapt the doses in routine practice: cisplatinum (I), methotrexate (47) and doxorubicin (36, 63, 64).

166

G. FREYER

From the theory

ET AL.

to the practice

Many oncologists are motivated to apply PK dose-adaptation strategies to patients receiving chemotherapy. Unfortunately, the ability of such methods is low, although the area of PWPD relationship has been extensively studied. In addition to the above-mentioned methodological limitations and problems, the low applicability of PWPD models in routine oncological practice is due to the necessity of performing multiple blood sampling in each new patient, except when a simple formula (such as Calvett’s or Chatelut’s) can predict the PK parameters on the basis of individual patient’s co-variables. The collaboration of the patient and the nurses, and the attention to rigorous infusion and sampling are required, a well as close collaboration with a PK unit. In the near future, the use of population data and specific software should enable the reduction of the number of required blood samples. Furthermore, oncologists might still be reluctant to use PK parameters because of the lack of randomized trials addressing the question of the usefulness of PK dose adaptation vs BSA dose adaptation, in terms of toxicity, tumour response and survival. Such trials are required to demonstrate, for each tumour model, the need of individual PK dose adaptation. In the authors’ opinion, potentially curative situations should be regarded as priorities for PK studies and PK dose adaptation, in order to reduce the clinical empirism that may compromise the individual benefit from chemotherapy. Highly chemosensitive tumours such as malignant lymphoma and germ-cell tumours might be particularly interesting from this point of view, as well as adjuvant post-operative treatments for breast or colorectal cancers, whose impact on the patient’s survival could well be improved.

Conclusion Pharmacokinetic studies should be encouraged in clinical oncology during Phase I trials, in order to early define the pharmacokinetic profile of drugs, as well as Phase II and III trials, in order to look for clinical and biological correlations. This descriptive step has to be performed prior to any further attempt of PK dose adaptation. Dose-adaptation control is obviously an elegant way to reduce inter-individual therapeutic variability. However, its real impact on treatment results has yet to be demonstrated in controlled trials in which standard chemotherapy regimens based on BSA dose adaptation should be chosen as control arms and compared to dose-adapted regimens based on PK approaches. Although associated with specific methodological and technical limits, PK studies and their clinical applications must aim towards the improvement of therapeutic results (65,661. Basic pharmacokinetic principles, excluding complex mathematical tools and models, should be made easily accessible to clinicians. Software and mathematical formulae should be made easier to use in routine practice, and more collaborative studies should be set up by clinicians and pharmacologists. Finally, one of the most promising applications of pharmacokinetics relies on the description of simple formulae derived from clinical trials and PK studies.

PHARMACOKINETIC

It could therefore become without actually petforming

STUDIES

IN CANCER

CHEMOTHERAPY

167

possible for clinicians to use pharmacokinetics pharmacokinetic studies.

References 1. Van Groeningen, C. J., Pinedo, H. M., Heddes, J. eta/. (1988) Pharmacokinetics of 5-fluorouracil assessed with sensitive mass spectrometric method in patients on a dose escalation schedule. Cancer Res. 48: 6956-61. 2. Albetto, P. (1995) Dose-intensity in cancer chemotherapy. Bull. Cancer82 (Suppl. 1): 3.~8~. 3. Gibaldi. M.. Levv. G. (1976) Pharmacokinetics in clinical oractice. J. Am. Med. Assoc. 235: 1864-67. ‘. 4. Benjamin, R. S., Wiernik, P. H. & Bachur, N. R. (1974) Adriamycin chemotherapy efficacy, safety and pharmacologic basis of an intermittent single high dosage schedule. Cancer 33: 19-27. 5. Tattersall, M. H. N., Parker, L. M., Pitman, S. W. & Frei, E. (1975) Clinical pharmacology of highdose methotrexate. Cancer Chemother. Rep. 6: 25-9. 6. Monfardini, S. & Chabner, B. (1991) Joint NCI-EORTC consensus meeting on neoplasia in the elderly. Eur. J. Cancer 27: 653-4. 7. Cockcroft, D. W. & Gault, M. N. (1976) Prediction of creatinine clearance from serum creatinine. Nephron. 16: 31-41. 8. Freyer, G., Maire, P., Ardiet, C., Tranchand, B. & Droz, J. P. (1995) Chemotherapy in the elderly: present and future. Bull. Cancer 82: 531-40. 9. Verotta, D., Beal, S. L. & Sheiner, L. B. (1989) Semiparametric approach to pharmacokineticpharmacodynamic data. Am. J. fhysiol. 256: 1005-10. 10. Milano, G., Etienne, M. G., Ramaiolo, A. eta/. (1994) Relationship between fluorouracil systemic exposure and tumor response and patient survival. J. C/in. Oncol. 12: 1291-5. 11. Bdrsi, J. D. & Moe, P. J..(1987) Prognostic importance of systemic clearance of methotrexate in childhood. Cancer Chemorher. Pharmacol. 19: 261-4. 12. Lind, M. J., Margison, J. M., Cerny, T. et a/. (1989) Prolongation of ifosfamide elimination halflife in obese patients due to altered drug distribution. Cancer Chemother. Pharmacol. 25: 139-42. 13. Nakamura, E., Moritani, T. & Kanetaka, A. (1989) Biological age versus physical fitness age. EUK J. Appl. Physiol. 56: 778-85. 14. Reece, P. H., Morris, R. G., Bishop, J. F. er a/. (1987) Creatinine clearance as a predictor of ultrafiltrable platinum disposition in cancer patients treated with Cisplatinum: relationship between peak ultrafiltrable platinum plasma levels and nephrotoxicity. J. C/in. Oncol. 5: 304-9. 15. Calve& A. H., Newell, D. R., Gumbrell, L. A. et a/. (1987) Carboplatin dosage: prospective evaluation of a simple formula based on renal function. J. C/in. Oncol. 7: 1748-56. 16. Chantler, C., Garnet& E. S., Parsons, V. et a/. (1969) Glomerular filtration rate measurement in man by single injection method using Cr-EDTA. C/in. Sci. 37: 169-180. 17. Jodrell, D. I., Egorin, M. J., Caretta, R. M. er a/. (1992) Relationship between carboplatin exposure and tumor response and toxicity in patients with ovarian cancer. J. C/in. Oncol. 10: 520-8. 18. Bleyer, W. A. (1978) The clinical pharmacology of methotrexate: new applications of an old drug. Cancer 41: 36-51. 19. Santini, J., Milano, G., Thyss, A. eta/. (1989) 5-FU therapeutic monitoring with dose adjustment leads to improved therapeutic index in head and neck cancer. f3r. J. Cancer 59: 287-90. 20. Sheiner, L. B. & Beal, S. L. (1981) Some suggestions for measuring predictive performance. J. Pharmacokinet. Biopharmaceut. 9: 503-12. 21. Hande, K. R., Wolff, S. N., Greco, F. A. eta/. (1990) Etoposide kinetics in patients with obstructive jaundice. J. C/in. Oncol. 6: 1101-7. 22. Desai, 2. R., Van der Berg, H. W., Bridges, J. M. & Shanks, R. G. (1982) Can severe vincristine neurotoxicity be prevented? Cancer Chemother: Pharmacol. 8: 211-14. 23. Launay, M. C., Milano, G., Iliadis, A. et a/. (1989) A limited sampling procedure for estimating adriamycin pharmacokinetics in cancer patients. Br. J. Cancer 60: 89-92. 24. Favre, R., Monjanel, S., Alfonsi, M. et a/. (1982) High-dose methotrexate: a clinical and pharmacokinetic evaluation. Cancer Chemotherz fharmacol. 9: 156-60. 25. Nirenberg, A., Mosende, C., Mehta, B. M. et a/. (1977) High-dose methotrexate with citrovorum factor rescus: predictive value of serum methotrexate concentrations and corrective measures to avert toxicity. Cancer Treat. Rep. 61: 77983. 26. Du Bois, D. & Du Bois, E. F. (1916) A formula to estimate the approximate surface area if height and weight to be known. Arch. Int. Med. 17: 863-71.

168 27. 28. 29. 30.

31. 32.

33.

34. 35. 36.

37.

38.

39.

40.

41. 42. 43. 44. 45. 46. 47. 48. 49. 50.

51. 52. 53.

G. FREYER

ETAL.

Balis, F. M. (1986) Pharmacokinetic drug interactions of commonly used anticancer drugs. C/in. Pharmacokinet. 11: 223-235. Evans, W. E. (1988) Clinical pharmacodynamics of anticancer drugs: a basis for extending the concept of dose intensity. Blut. 56: 241-8. Rodvold, K. A., Rushing, D. A. & Tewksbury, D. A. (1988) Doxorubicin clearance in the obese. J. C/in. Oncol. 6: 1321-7. EORTC Pharmacokinetics and metabolism group. (1987) Pharmacokinetically guided dose escalation in phase I clinical trials. Commentary and proposed guidelines. Eur. J. Cancer C/in. Oncol. 23: 1083-7. Schilsky, R. (1982) Renal and metabolic toxicities of cancer chemotherapy. Sem. Oncol. 9: 75-83. Egorin, M. J. (1992) Therapeutic drug monitoring and dose optimization in Oncology. In: Workman, P., ed. New Approaches in Cancer Pharmacology: Drug Design and Development. Springer, Berlin, Heidelberg, New York, pp. 75-91. Van Warmerdam, L. Van den Bemt, B., Ten Bokkel Huinink, W. eta/. (1995) Dose individualisation in cancer chemotherapy: pharmacokinetic and pharmacodynamic relationships. Cancer Res. Ther. Control 4: 277-91. Extra, J. M., Rousseau, F., Bruno, R. et a/. (1993) Phase I and pharmacokinetic study of Taxotere given as a short intravenous infusion. Cancer Res. 63: 1037-42. Kerr, I. G., Jolivet, J., Collins, J. M. eta/. (1983) Test-dose for predicting high-dose methotrexate infusions. C/in. Pharmacol. Ther: 33: 44-51. Launay-lliadis, M. C., Bruno, R., Montay, G. eta/. (1993) Population pharmacokinetics of Taxotere using non-parametric maximum likelihood (NPML) software. Eur. J. Drug Metab. Pharmacokinet. 18: 196 (Abst). Bruno, R., Dorr, M. B., Montay, G. et a/. (1994) Design and prospective implementation of population pharmacokinetic studies during the development of Docetaxel, a new anticancer drug. C/in. Pharmacol. Ther. 55: 2-35. Kawato, Y., Aonuma, M., Hirato, Y. et a/. (1992) Intracellular roles of SN-38, a metabolite of the camptothecin derivative CPT-11 in the antitumor effect of CPT-42. Biochem. Biophys. Res. Commun. 188: 70-7. Catimel, G., Chabot, G. G., Guastalla, J. P. et a/. (1995) Phase I and pharmacokinetic study of lrinotecan (CPT-11) administered daily for 3 days every 3 weeks in patients with advanced solid tumors. Ann. Oncol. 6: 133-40. Chabot, G. G., Abigerges, D., Catimel, D. et al. (1995) Population pharmacokinetics and pharmacodynamics of lrinotecan (CPT-11) and active metabolite SN-38 during phase I trials. Ann. Oncol. 6: 141-151. de Forni, M., Bugat, R., Chabot, G. G. et a/. (1993) Phase I and pharmacokinetic trial of weekly CPT-11. J. C/in. Oncol. 11: 2194-2204. Canal, P., Gay, C., Dezeuze, A. et a/. (1996) Pharmacokinetics and pharmacodynamics of lrinotecan during a phase II clinical trial in colorectal cancer. J. C/in. Oncol. 14: 2688-95. Eksborg, S. (1989) Anthracycline pharmacokinetics. Limited sampling model for plasma level monitoring with special reference to epirubicin. Acta Oncol. 29: 339-42. Ratain, M. J., Robert, J. &Van der Vigh, W. F. J. (1991) Limited sampling models for doxorubicin pharmacokinetics. J. C/in. Oncol. 9: 871-6. Hryniuk, W. M. (1986) Analysis of dose-intensity for adjuvant chemotherapy trials in stage II breast carcinoma. JCO 4: 65-74. Lena, N. (1987) Methotrexate test-dose protocol in the presence of 7-hydroxymethotrexate. Eur. J. Cancer Clin. Oncol. 23: 481-485. Iliadis, A. & Bachir-Raho, M. (1985) Bayesian estimation and prediction of clearance in highdose methotrexate infusions. J. Pharmacokin. Biopharm. 13: 101-115. Newell, H., Calvert, H., Balmanno, K. et a/. (1990) Carboplatin pharmacokinetics in children. Winter Meeting of the EORTC-PAM Group, Bordeaux, November. Chatelut, E., Canal, P., Brunner, V. eta/. (1995) Prediction of carboplatin clearance from standard morphological and biological patient characteristics. J. Nat/. Cancer Inst. 87: 573-80. Ghazal-Aswad, S., Calvert, A. H.&Newell, D. R. (1996) A single-sample assay for the estimation of the area under the free carboplatin olasma concentration versus time curve. Cancer Chemother: Pharmacol. 37: 429-34. . Hayder, S., Lafolie, P., Bjorko, 0. &Peterson, C. (1989) 6-Mercaptopurin plasma levels in children with acute leukemia: relation to relapse risk and mvelotoxicitv. Ther. Drug Monit. 11: 617-22. Desoize, B., Marechal, F. & Cattan, A: (1990),Clinicaipharmacdkinetics of etoposide during 120 hours continuous infusions in solid tumors. Br. J. Cancer 62: 840-I. Piscitelli, S. C., Rodvold, K. A., Rushing, D. A. & Tewksburry, D. A. (1993) Pharmacokinetics and pharmacodynamics of doxorubicin in patients with small cell lung cancer. C/in. Pharmacol. Ther. 53: 555-61.

PHARMACOKINETIC 54. 55. 56.

57.

58. 59.

60. 61.

62.

63. 64.

65. 66. 67. 68. 69.

70. 71. 72.

73.

STUDIES

IN CANCER

CHEMOTHERAPY

169

Evans, W. E., Crom, W. R., Abromovirch, M. et al. (1986) Clinical pharmacodynamics of highdose methotrexate in acute lymphocytic leukemia. New Engl. J. Med. 314: 471-7. Stewart, C. F., Arbuck, S. G., Fleming, R. A. 81 Evans, W. E. (1991) Relation of systemic exposure to unbound etoposide and hematologica toxicity. C/in. fbarmacol. Ther: 50: 385-93. Rustum, Y. M., Riva, R. C. & Preisler, H. D. (1987) Pharmacokinetic parameters of Ara-C and their relationship to intracellular metabolism of Ara-C, toxicity and response of patients with acute non lymphocytic leukemia treated with conventional and high-dose Ara-C. Sem. Dncol. 14: 141-8. Longnecker, S. M., Donehower, R. C., Cates, A. E. et a/. (1987) High-performance liquid chromatographic assay for taxol in human plasma and urine and pharmacokinetics in a phase I trial. Cancer Treat Rep. 71: 53-9. Gamelin, E. (1994) Intensified chemotherapy of metastatic colorectal cancer with individual adaptation of 5-fluorouracil dose. froc. Am. Assoc. Cancer Res. 35: 411. Kearns, C., Gianni, L., Capri, G., Gianni, A. & Egorin, M. (1994) A comprehensive pharmacokinetic (PK) model of paclitaxel and 6-OH paclitaxel with application to pharmacodynamic (PD) relationship in humans. Proc. Am. Sot. C/in. &co/. 13: 134. Le Blanc, G. & Waxman, D. (1989) Interaction of anticancer drugs with hepatic monooxygenase enzymes. Drug Met. Rev. 20: 395-439. Jelliffe, R. W., Schumitzky, A. et al. (1993) Individualizing drug dosage regimens: role of population pharmacokinetic and dynamic models, bayesian fitting and adaptative control. Ther. Drug Monitoring 15: 380-93. Boeckmann, A. J., Sheiner, L. B. & Beal, S. L. (1992) NONMEM users guides, part V: introductory guide. Technical Report of the Division of Clinical Pharmacology, University of California. San Francisco: University of California. Bressolle, F., Ray, P., Jacquet, J. M. et al. (1991) Bayesian estimation of doxorubicin pharmacokinetic parameters. Cancer Chemother. fharmacol. 29: 53-60. Ratain, M. J., Mick, R., Schilsky, R. L., Vogelzang, N. J. & Berezin, F. (1991) Pharmacologically based dosing of etoposide: a means of safely increasing dose intensity. J. C/in. Oncol. 9: 1480-6. Collins, J. M., Zaharko, D. S., Dedrick, R. L. & Chabner, B.A. (1986) Potential roles for preclinical pharmacology in phase I clinical trials. Cancer Treat Rep. 70: 73-80. Collins, J. M. (1988) Improving the use of anticancer drugs: clinical pharmacokinetic approaches. Is. J. Med. Sci. 24: 483-7. Egorin, M. J., Echo van D. A., Tipping, S. J. eta/. (1984) Pharmacokinetics and dosage reduction of cisplatinum in patients with impaired renal function. Cancer Res. 44: 5432-8. Ratain, M. J., Schilsky, R. L., Conley, B. A. 81 Egorin, M. J. (1990) Pharmacodynamics in cancer therapy. J. C/in. Oncol. 8: 1739-53. Sonnichsen, D., Ribeiro, R., Luo, X. et a/. (1994) Pharmacokinetics and pharmacodynamics of 21-day continuous oral etoposide (VP-161 in pediatric solid tumor patients. Proc. Am. Sot. C/in. Oncol. 13: 140. Stoller, R. G., Hande, K. R., Jacobs, S. A. et a/. (1977) Use of plasma pharmacokinetics to predict and prevent methotrexate toxicity. N. Engl. J. Med. 297: 630-4. Tranchand, B., Point, Y. D., Minuit, M. P. et al. High-dose melphalan adjustment: possibility of using a test-dose. Cancer Chemother. 23: 95-100. Ackland, S. P., Ratain, M. J., Vogelzang, N. J. et al. (1989) Pharmacokinetics and pharmacodynamics of long-term continuous-infusion Doxorubicin. C/in. Pharmacol. Ther. 45: 340-347. Pinkel, D. (1958) The use of body surface area as a criterion of drug dosage in cancer chemotherapy. Cancer Res. 18: 853-6.