INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Recombinant human granulocyte colony-stimulating factor (rhG-CSF) is widely used for patients with neutropenia during cancer chemotherapy. The pharmacokinetics of rhG-CSF, however, are not fully understood, and the optimal schedule for rhG-CSF treatment remains undetermined. Previous studies, including those from our laboratories, have reported that serum rhG-CSF levels inversely correlate with the number of circulating neutrophils in cancer patients (20-22) and that rhG-CSF is possibly eliminated through G-CSF receptors on neutrophils (7, 24). In contrast, Kearns et al. (12) reported that low levels of rhG-CSF clearance (CL_G-CSF) remained stable in patients with severe neutropenia. This finding suggests that another mechanism, apart from circulating neutrophils, is involved in the clearance of rhG-CSF. CL_G-CSF is low in rodents with renal failure (11, 17, 23), and the kidney may account for the alternative clearance mechanism of rhG-CSF.

Since anticancer drugs, such as platinum agents, high doses of methotrexate, and nitrosoureas, induce both nephrotoxicity and neutropenia, it is important to clarify the effects of renal function on the pharmacokinetics of rhG-CSF. However, to our knowledge, there are no studies that have examined the relationship between renal function and the pharmacokinetics of rhG-CSF in cancer patients receiving chemotherapy. The subjects of our previous studies were patients with normal renal function, and thus the influence on renal function could not be evaluated (22, 24). The present study included lung cancer patients with variable renal function. Using multivariate analysis, we demonstrate here that CL_G-CSF is reduced in lung cancer patients with poor renal function.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Patient selection. The present study was conducted according to institutional ethical standards. Patients were consecutively selected if they fulfilled the following criteria: (i) the presence of histologically or cytologically confirmed lung cancer; (ii) an Eastern Cooperative Oncology Group performance status of 2 or better; (iii) eligibility for chemotherapy; (iv) no prior chemotherapy or radiotherapy; (v) absence of bone metastasis; (vi) adequate bone marrow function with absolute neutrophil counts (ANC) of >2,000/µl, platelet counts of >100,000/µ, and hemoglobin levels of >10 g/dl; (vii) normal hepatic function; and (viii) informed consent of the patient for the study. Twenty-five patients with lung cancer were consecutively enrolled in the study. All tests and analytical procedures were performed during the first course of chemotherapy.

rhG-CSF administration and sample collection. The counts of circulating leukocytes or neutrophils were monitored three times weekly for the detection of chemotherapy-induced leukopenia or neutropenia. A bolus dose of 5 µg of rhG-CSF (Lenograstim; Chugai Pharmaceutical Co., Tokyo, Japan)/kg of body weight was intravenously injected at 9 a.m. on the day on which the ANC <1,000/µl or leukocyte counts were <2,000/µl until the first day on which the ANC>5,000/µl or leukocyte counts were >10,000/µl. For the 25 patients enrolled in the study, pharmacokinetic studies were conducted on the first day of rhG-CSF administration for 20 patients and on the last day of rhG-CSF administration for 5 patients. Blood samples for the ANC and serum creatinine were obtained just before rhG-CSF administration on the same day of the pharmacokinetic study. Blood samples for pharmacokinetic study were collected just before rhG-CSF administration and at at least three other points until 8 h after administration. A total of 125 points after administration were collected from 25 patients, and the mean number of concentrations measured per patient was 5 points (range, 3 to 7 points). The sera were immediately centrifuged and stored at 20°C until they were assayed.

Serum G-CSF concentration. Serum G-CSF levels were measured by a chemiluminescence enzyme immunoassay (15). In this assay, serum G-CSF is sandwiched between anti-G-CSF immunoglobulin G-coated beads and glucose oxidase-labeled antibody. Supplementation of glucose results in the production of hydrogen peroxide through the action of glucose oxidase. Production of H₂O₂ is determined chemiluminescently with a mixture of luminol and potassium ferricyanide. This chemiluminescence assay is highly sensitive and detects as little as 2 pg of G-CSF/ml in the serum (15). The reproducibility of this assay was confirmed with intra- and interassay coefficients; the variance ranged from 5.5 to 7.8 and 3.4 to 16.0%, respectively (2). Serum rhG-CSF concentrations were calculated by subtracting the G-CSF concentration before administration from the G-CSF concentration obtained.

Renal function. Serum creatinine levels were determined by method of the Jaffe (9), and creatinine clearance (CL_CR) was predicted by the Cockcroft-Gault formula (5). The glomerular filtration rate estimated by this formula could be overestimated, since serum creatinine depends on muscle mass, which may be reduced in patients with lung cancer. However, the formula has been used in several studies of cancer patients receiving platinum-based chemotherapy, and it shows a good correlation between CL_CR and the glomerular filtration rate (8, 19). In comparison, estimation of 24-h CL_CR is a time-consuming procedure and is often unreliable due to incomplete 24-h urine collection (6). The Cockcroft-Gault formula is practical for rapid evaluation of renal function.

Pharmacokinetic analysis. Population pharmacokinetics were performed using the nonlinear mixed-effect model (NONMEM) program (version V; University of California at San Francisco, San Francisco) (4). As the analysis model, we estimated one-compartment and two-compartment models with bolus input and first-order output. The appropriateness of the fitting model was determined by the objective function. The parameters selected to describe the model were clearance (CL) and distribution volume (V) for the one-compartment model and CL, volume of the central compartment (V₁), volume of the peripheral compartment (V₂), and intercompartmental clearance (CL_int) for the two-compartment model. The interindividual variability was assumed to be log-normally distributed as follows:
where P_i is the value of the parameter of the individual, _i,tv is the typical value (tv) of this parameter in the population, and  is a variable accounting for interindividual variability, with a mean of zero and variance ². The residual error, , was also assumed to be log-normally distributed with a mean of zero and variance ² as follows:
where Y is the dependent variable (i.e., plasma concentration), which is a function of the known quantity x (i.e., time) and the pharmacokinetic parameter . The pharmacokinetic parameters of individual patients were calculated by the Bayesian method. The area under the concentration-time curve (AUC) was calculated by dividing the actual dose of rhG-CSF by CL_G-CSF, and the elimination constant rate (K_e) was determined by the expression CL_G-CSF/V or CL_G-CSF/V₁.

Statistical analysis. The values of the ANC were log transformed for normalization. As univariate analysis, the Pearson correlation coefficient (r) was calculated to evaluate the relationship between CL_G-CSF and clinical factors. Furthermore, multiple linear regression was conducted as multivariate analysis (13). A final set of the significant factors was selected in a stepwise fashion. The coefficient of determination (R²) was used to assess the variability in CL_G-CSF that the combination of these factors accounted for. The contribution of each factor to the variability was calculated by multiplying the standard regression coefficient by the r value. A two-tailed P of <0.05 was considered significant. Data were analyzed with the StatView software program (version 5.0; SAS Institute Inc., Cary, N.C.).

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Twenty patients were males, and five were females, with a median age of 63 years (range, 35 to 79 years). The histologic type of lung cancer was adenocarcinoma in 6 patients, squamous cell carcinoma in 4, and small-cell carcinoma in 15. Three tumors were stage I disease, 8 were stage IIIA, 3 were stage IIIB, and 11 were stage IV. All chemotherapeutic regimens used for the treatment of these patients contained platinum agents, including cisplatin for 8 patients and carboplatin for 17. The mean pretreatment level of CL_CR for the entire group was 65.6 ± 22.0 ml/min, and the median, 25th-percentile, and 75th-percentile values of the ANC at nadir were 651, 530, and 875/µl, respectively.

The population pharmacokinetic parameters of rhG-CSF were calculated by using the NONMEM program. The two-compartment model was found to describe the data better than the one-compartment model (data not shown). The population mean values were 0.65 liters/h for CL_G-CSF, 3.09 liters for V₁, 497 liters for V₂, and 0.10 liters/h for CL_int. The interindividual variability was 51.5% for CL_G-CSF and 29.5% for V₁. The residual intraindividual variability was 13.4%.

Individual pharmacokinetic parameters were estimated by the Bayesian method using the above population parameters. The means (±standard deviation) of individual parameters were as follows: CL_G-CSF, 0.83 ± 0.39 liters/h; AUC, 408 ± 201 ng · h/ml; V₁, 3.01 ± 0.82 liters; and K_e; 0.27 ± 0.11/h. Table 1 shows the CL_G-CSF levels and clinical characteristics of participating patients on the day of rhG-CSF administration. In univariate analysis CL_G-CSF correlated significantly with the ANC (r = 0.613) and CL_CR (r = 0.632) (Fig. 1) but not with albumin levels, liver function, or other hematological factors. Multiple linear regression analysis showed that the ANC and CL_CR were significant variables among the clinical factors and that the combination of the ANC and CL_CR accounted for 57.4% of the variation of CL_G-CSF (Fig. 2). As a whole, the ANC contributed 27.4% of CL_G-CSF and CL_CR was responsible for 30.0% of CL_G-CSF. However, CL_CR accounted for 72.9% of CLG-_CSF in patients with low ANC (<1,000/µl) (Fig. 3).

View larger version (15K): [in this window] [in a new window]   FIG. 1.   Correlation between rhG-CSF clearance and circulating neutrophil counts (top) and creatinine clearance (bottom).

View larger version (19K): [in this window] [in a new window]   FIG. 2.   Relationship between actual rhG-CSF clearance and estimated rhG-CSF clearance based on the multiple linear regression equation CLG-CSF (liter/h) = 0.008 CLCR (ml/min) + 0.352 log ANC (per µl)  0.809.

View larger version (13K): [in this window] [in a new window]   FIG. 3.   Relationship between rhG-CSF clearance and creatinine clearance in patients with neutrophil counts of <1,000/µl.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

The present study demonstrated that CL_G-CSF levels correlated with renal function as well as the number of circulating neutrophils in patients with lung cancer receiving chemotherapy. The combination of CL_CR and the ANC accounted for approximately half of the CL_G-CSF variation. As a whole, the contribution of CL_CR to CL_G-CSF was almost equal to that of the ANC; however, the contribution of CL_CR became larger in lung cancer patients with neutropenia.

Our clinical observation supported the previous findings in rodent models demonstrating a reduction in CL_G-CSF following ligation of the renal artery or heminephrectomy (11, 23). In addition, a high dose of rhG-CSF in rats saturates the clearance mechanism of G-CSF receptor-positive cells, and further ligation of the renal artery decreases this saturated clearance (17). In humans, Akizawa et al. (1) reported that the elimination half-life and AUC of rhG-CSF seems to be increased in patients with chronic renal failure, although their study did not include control subjects. These findings suggest that the kidney is involved in the elimination of rhG-CSF through a nonsaturable process. In support of this conclusion, other studies have demonstrated the preferential accumulation of radiolabeled rhG-CSF in the rat kidney just after intravenous administration (14, 18). The radioactivity in the urine is 59% of the total dose (14), but less than 1% of rhG-CSF is detected by enzyme-linked immunosorbent assay (ELISA) (23). Thus, the elimination of rhG-CSF is possibly due to metabolism by the kidney.

As a whole, the contribution of renal function to CL_G-CSF was almost equal to the influence of circulating neutrophil counts. The clearance process by G-CSF receptor-positive cells accounts for approximately 80% of CL_G-CSF in healthy subjects (18). In patients with relatively adequate neutrophil counts, neutrophils are mainly involved in the elimination of rhG-CSF. Kearns et al. (12), however, reported that CL_G-CSF decreased as the ANC diminished from 100,000 to 1,000/µl but that CL_G-CSF remained stable in patients with neutropenia of <1,000/µl. In the present study, CL_CR accounted for as much as 72.9% of the variation of CL_G-CSF in patients with neutropenia (counts, <1,000/µl). Thus, the overall effect of renal function on CL_G-CSF was almost equal to that of neutrophils, but renal function played a major role in CL_G-CSF in lung cancer patients with neutropenia.

The combination of CL_CR and the ANC accounted for half of the CL_G-CSF variation, and the equation obtained from multiple regression analysis was not completely predictive. These results suggest the presence of other clearance mechanisms. G-CSF receptors are involved in the saturable clearance of rhG-CSF (18, 24), and they are expressed on cells other than circulating neutrophils, such as bone marrow myeloid cells (3). In this regard, the elimination half-life of rhG-CSF in patients with myelodysplastic syndrome and aplastic anemia correlates with the number of bone marrow myeloid cells (25). Furthermore, soluble G-CSF receptors are released from mature myelomonocytic cells to the serum, and the serum levels correlate with the number of circulating neutrophils and monocytes during rhG-CSF therapy (10). These findings indicate that the pharmacokinetics of rhG-CSF are probably modulated by cellular and soluble G-CSF receptors in a complex fashion.

Other nonsaturable mechanisms may also exist, since the kidney is responsible for 40 to 50% of the nonsaturable clearance of rhG-CSF, as demonstrated previously in a rat model (17). The liver can also metabolize many substances; however, studies using radiolabeled rhG-CSF showed accumulation of only small amounts in the liver (14, 18). In partially hepatectomized rats, the decline in CL_G-CSF is small, and the change is due to the volume of distribution (23). In the present study, liver function did not correlate with CL_G-CSF. Accordingly, the liver does not seem to be involved in the nonsaturable clearance of rhG-CSF. Further studies are necessary to clarify all those factors that could affect the pharmacokinetics of rhG-CSF in cancer patients.

The values of K_e (0.27/h) and V₁ (3.01 liters) in the present study were similar to those of previous studies of healthy and neutropenic subjects (16, 25). In intravenous rhG-CSF administration, the K_e and V₁ values were 0.32/h and 4.56 liters, respectively, for 60-kg healthy volunteers (16) and 0.26/h and 2.88 liters for 60-kg patients with aplastic anemia (25). In contrast, K_e and V₁/bioavailability in healthy volunteers injected subcutaneously with rhG-CSF were 0.220 to 0.288/h and 14.5 liters, respectively (7). The high V₁/bioavailability values were considered to be due to subcutaneous administration. For the determination of serum G-CSF levels, chemiluminescence enzyme immunoassay was used in the present study, whereas ELISA was used in previous studies (7, 16, 25). The use of different assays, however, is unlikely to have influenced the results of the present study, since chemiluminescence enzyme immunoassay correlates well with ELISA (15).

In conclusion, we demonstrated that renal function as well as the number of circulating neutrophils correlated with CL_G-CSF and that renal function played an important role in the elimination of rhG-CSF in lung cancer patients with neutropenia. These findings suggest that, in lung cancer patients scheduled for rhG-CSF therapy, the dosage of rhG-CSF could be adjusted according to the ANC and CL_CR on the day of administration.