Gentamicin Pharmacokinetics in Patients with Malignancies

Vol. 35, No. 7

ANTIMICROBIAL AGENTS AND CHEMOTHERAPY, July 1991, p. 1501-1503

0066-4804/91/071501-03$02.0010 Copyright © 1991, American Society for Microbiology

Gentamicin Pharmacokinetics in Patients with Malignancies JOSEPH S. BERTINO JR. ,12* LEIGH ANN BOOKER,' PATRICK FRANCK,' AND BENJAMIN RYBICKI3 Departments of Pharmacy Services' and Medicine2 and Research Institute,3 The Mary Imogene Bassett Hospital, Cooperstown, New York 13326 Received

5

November 1990/Accepted 15 April 1991

The pharmacokinetics of gentamicin were investigated in 880 patients with leukemia (24 patients), other malignancies (211 patients), or no malignancies (645 patients) by using data collected by our Clinical Pharmacy Service. A significant difference was seen in the initial calculated creatinine clearance between the patients with leukemia and the other two groups. No differences in gentamicin pharmacokinetics were seen in patients with other malignancies versus those with no malignancies. Patients with leukemia had significantly faster drug clearance compared with those in the other two groups. A poor predictive value was found for total body clearance of gentamicin versus the initial calculated creatinine clearance in all groups. Multiple logistic regression analysis showed that only the initial calculated creatinine clearance differed in the leukemic group compared with those in the other patients. Our data suggest that no pharmacokinetic difference exists for gentamicin in patients with malignancies.

Aminoglycoside antibiotics remain a valuable therapeutic class in patients with malignancies who become febrile with a suspected bacterial infection (2). Pharmacokinetic monitoring and appropriate dosage adjustment for these agents results in improved patient outcome in both neutropenic and nonneutropenic patients (9, 10, 14). Previous data have suggested that the pharmacokinetics of aminoglycoside antibiotics may differ in patients with malignancies (6, 8, 11, 17). The purpose of our investigation was to examine prospectively collected data on gentamicin pharmacokinetics in the following three adult patient (over 18 years old) populations: (i) patients with acute leukemia, (ii) patients with other nonleukemic malignancies, and (iii) patients with no underlying malignancy or pathophysiology other than renal impairment known to alter gentamicin pharmacokinetics. This study was done with data collected prospectively (and analyzed retrospectively) by the Clinical Pharmacy Service of The Mary Imogene Bassett Hospital from January 1983 through January 1989. Patients eligible for the study included those with leukemia, those with other malignancies (including lymphomas), and those who received gentamicin but who had no pathophysiology (aside from impaired renal function) known to alter gentamicin pharmacokinetics (3, 7, 15, 16). The data collected included age, sex, total body weight, lean body weight, initial serum creatinine level, and initial calculated creatinine clearance (CLCR) (by the method of Cockcroft and Gault [4]). Gentamicin pharmacokinetic parameters were determined by the method of Sawchuck and Zaske (13) by using a 30-min infusion and a three- or four-point determination. Data were fit to a one-compartment intravenous infusion, pharmacokinetic model. Predose blood samples for serum gentamicin concentration analysis were obtained, with two or three additional samples obtained after the infusion. Serum gentamicin concentrations were adjusted to obtain 1-h peak concentrations (1 h after the beginning of the infusion) of 5 to 10 ,ug/ml (7 to 10 p,g/ml for patients with pneumonia) and trough concentrations of less than 2 p,g/ml, regardless of the dosing interval. Serum gentamicin concentrations were determined by the *

fluorescence polarization immunoassay technique. The interim coefficient of variations were 2.2, 1.5, and 0.9% at 1, 4, and 8 ,ug/ml, respectively. The data were analyzed on a Microvax II computer by using Version 5 of the SAS (SAS Institute) (12). Betweengroup differences in discrete variables were examined by using the Pearson chi-square statistic and the Fisher exact test when appropriate. Differences in continuous variables between groups were examined by using the Student t test for unpaired observations when comparing two groups and analysis of variance when comparing all three groups. Both the Duncan multiple range and the Scheffe tests were used to test for individual group differences when a statistically significant variable resulted from the analysis of variance. The variables that were significant on a univariate level were put into a multiple logistic regression equation to test for the independence of each variable's effect on the outcome. All continuous variables are expressed as means + standard deviations. A total of 880 patients were monitored over the specified time period. This included 24 patients with leukemia (Table 1), 211 patients with other malignancies (Table 1), and 645 patients with no malignancies. Table 2 illustrates the demographics and pharmacokinetics for four groups of patients: the entire group (the data for this group are reported for reference only), leukemic patients only, patients with other malignancies only, and patients with no malignancies. No significant difference was seen in the volume of distribution at steady state (V.,) (liters per kilogram of total body weight) between any groups. There was no significant difference in half-life (t4/2) or total body clearance (CL) between the group with malignancies (other than leukemia) versus the group with no malignancy. No significant difference in the initial calculated CLCR was noted between these two groups. Selected variables of the leukemic, malignancy, and no malignancy groups were compared. This comparison resulted in an initial calculated CLCR which was statistically significantly different (P < 0.05) and a CL that was statistically significantly different (P = 0.05). These differences in CLCR and CL between the leukemic and the other two groups were further examined in a multiple logistic regression model that controlled for Vss and tl/2. Leukemia was the

Corresponding author. 1501

1502

NOTES

ANTIMICROB. AGENTS CHEMOTHER.

TABLE 1. Data on 235 patients with malignancy

TABLE 3. Multiple logistic regression model for pharmacokinetic variablesa

No. of patients ptet

Malignancy Leukemiasa ANLL ............................................. ALL ............................................. Other malignancies (by primary major organ site)

Lung .............................................. Non-Hodgkin's Lymphoma .......................................... Colon ............................................. Pancreas ............................................. Breast ............................................. Prostrate ............................................. Bladder ............................................. Rectal ............................................. Oral cavity .............................................. Female genitourinary .............................................. Liver .............................................

Kidney ............................................. Brain .............................................

Gastric ............................................. Soft Tissue .............................................

Billiary ............................................. Skin .............................................. Adrenal ............................................. Testicular .............................................. Unknown primary .....................................

.........

Variable

Chi-square

P value

19 5

Intercept Calculated CLCR (ml/min/1.73 m2) Gentamicin CL (ml/min/1.73 m2)

53 37 26 14 13 12 9 8 6 4 4 4 4 3 3 2 2 1 1 5

t112 (h)

15.28 5.62 0.07 0.21 0.06

0.0001 0.018 0.798 0.646 0.81

a ANLL, acute nonlymphocytic leukemia; ALL, acute lymphocytic leukemia.

dependent variable in this model. As shown in Table 3, with all four variables present in the model (CLCR, CL, V,r, and t4/2), only CLCR was statistically significantly different (P < 0.05). Finally, correlation coefficients to determine the ability of CLCR to predict gentamicin CL were generated from weighted least-squares linear regressions. Four linear regressions were done: one for each group and one for the entire group. The equations that resulted are shown in Table 4. It is interesting to note the marked difference in the linear relationship of gentamicin CL to CLCR in the leukemic group compared with those in the other groups. All four correlation coefficients were statistically significant; however, as we have previously illustrated in febrile neutropenic patients (1), their predictive ability for clinical use is poor. A number of studies have reported alterations in the pharmacokinetics of aminoglycosides in patients with malig-

V, (liter/kg) a The dependent variable was leukemia.

nancies (6, 8, 11). The studies of Manny and Hutson (8) and Higa and Murray (6) assumed a control population V,s of 0.25 to 0.3 liter/kg with no matched controls. The study of Phillips et al. (11) compared aminoglycoside pharmacokinetics in febrile neutropenic cancer patients, but matched subjects only for age and sex but not for renal function or weight. More recently, Zeitany et al. (17) reported significantly different Vss, t1/2, and CL values for aminoglycosides in patients with hematologic malignancies compared with those in control patients. Unfortunately, they did not report comparative CLCR data, and thus, it is difficult to assess whether a difference in renal function could explain the differences in t1/2 and CL. Our study revealed no difference in gentamicin pharmacokinetic parameters between patients with various malignancies versus those with no malignancies. Patients with leukemia did have higher initial calculated CLCR values compared with those in the group with or without malignancies. This difference in initial calculated CLCR in the group with leukemia resulted in a significantly higher gentamicin CL in this group, which is not unexpected in patients with superior renal function. When multiple logistic regression was performed on the data, however, only initial calculated CLCR remained statistically different in the leukemic group compared with those in the other two groups. A poor predictive value was seen for CL when the initial CLCR was used in a regression equation for all groups. In summary, we reported gentamicin pharmacokinetics in four patient groups. No significant difference was found in pharmacokinetic parameters between patients with malignancies versus those with no malignancies. A significant difference was noted in the initial calculated CLCR for patients with leukemia. A poor predictive value was noted for CL when the initial CLCR was used. Individualized

TABLE 2. Demographics and pharmocokinetics of the three groups of patientsa Gender distribution

Group

Total (n

=

880)

Leukemia (n

=

Malignancy (n

24) =

211)

No malignancy (n

=

655)

(no.)' 456 M 424 F 12 M 12 F 107 M 104 F 337 M 308 F

Age (yr)

Idealbody

Ideal(k) t(g

Total body wt (kg) t(g

Initial calculated CLCR (ml/min/ 1.73 M2)d

V, (liter/kg

of total body

t1/2 (h)

G

CL

(ml/min/1.73 M2)

wt)(mmiI73n)

61.0 ± 16.9 62.4 ± 11.3 71.9 ± 19.5

60.3 ± 28.9

0.34 ± 0.11 4.7 ± 6.1

75.9 ± 39.6

54.2 ± 18.2 62.8 ± 11.6 74.5 ± 14.7

80.8 ± 38.le

0.34 ± 0.11 3.4 ± 2.3

91.8 ± 34.8e

65.1 ± 12.5 62.2 ± 11.0 68.7 ± 17.2

59.1 ± 25.6

0.35 ± 0.12 4.5 ± 3.6

75.7 ± 38.6

60.3 ± 18.2 62.2 ± 11.1 72.9 ± 19.8

59.8 ± 29.9

0.33 ± 0.11 4.8 ± 6.8

75.3 ± 39.7

a Values are means ± standard deviations. b M, male; F, female. c Calculated by the method of Devine (5). d Calculated by the method of Cockcroft and Gault (4). P p 0.05 from the other two groups.

NOTES

VOL. 35, 1991

TABLE 4. Linear regression equations for CL versus initial calculated CLCR Regression equation

Group

CL CL Leukemic patients CL Malignancy patients CL

Total No malignancy a

= = = =

25.3 24.2 18.5 30.1

+ + + +

0.47 (CLCR) 0.47 (CLCR) 0.68 (CLCR) 0.4 (CLCR)

P value

Predictive

'0-001 '0-001 '0-002