Effect of Protein Binding in Serum on Therapeutic Efficacy of Cephem

Home , Cefpiramide

ANTIMICROBIAL AGENTS AND CHEMOTHERAPY, Jan. 1992, p. 17-24 Vol. 36, No. 1 0066-4804/92/010017-08$02.00/0 Copyright © 1992, American Society for Microbiology Effect of Protein Binding in Serum on Therapeutic Efficacy of Cephem Antibiotics SHUICHI TAWARA,1 SATORU MATSUMOTO,' TOSHIAKI KAMIMURA,1* AND SACHIKO GOTO2 New Drug Research Laboratories, Fujisawa Pharmaceutical Co., Ltd., 2-1-6 Kashima, Yodogawa-ku, Osaka 532,1 and Department ofMicrobiology, Toho University School ofMedicine, 5-21-16 Ohmori Nishi, Ohta-ku, Tokyo 143,2 Japan Received 29 May 1991/Accepted 2 October 1991 The effect of protein binding in serum of eight cephem antibiotics (ceftazidime, ceftizoxime, cefotiam, cefmetazole, cefpiramide, cefazolin, cefuzonam, ceftriaxone) on their therapeutic efficacies was examined in mice with experimentally induced intraperitoneal infections or pneumonia. The relationship among therapeutic activity, in vitro antibacterial activity, total or free (unbound) levels in serum, and homogenized whole lung levels was investigated. In the intraperitoneal infection caused by Staphylococcus aureus or Klebsiella pneumoniae, the 50% effective doses (ED50s) of the cephem antibiotics correlated with the area under the concentration-time curve (AUC) values of free levels in serum and the MICs but not with those of total levels in serum. A linear relationship was seen between V/ED.0 values and AUC of free levels in serum/MIC valaues. On the other hand, in mice with pneumonia caused by K. pneumoniae, the number of bacteria in the lung closely correlated with the AUC of the antibiotic concentration in lung tissue. There was a direct correlation between the levels in lung tissue and total levels in serum but not free levels in serum. The cephem antibiotics tested in this study were bound only slightly to homogenates of mouse lung. These results indicate that the effect of protein binding in serum on therapeutic efficacy against intraperitoneal infection differs from that against pulmonary infection.

The significance of antibiotic protein binding has been MATERIALS AND METHODS discussed for the past 40 years. Protein binding has two Antimicrobial agents. The following antimicrobial agents direct effects on therapeutic efficacy because it affects anti- were used: cefazolin (Fujisawa Pharmaceutical Co., Ltd.), bacterial activity and distribution. It is accepted that only the cefotiam (Takeda Chemical Industries, Ltd.), cefmetazole free fraction can act against bacteria (23, 29). The effect of (Sankyo Co., Ltd.), ceftizoxime (Fujisawa Pharmaceutical protein binding in serum on extravascular penetration is less Co., Ltd.), ceftriaxone (Hoffman-LaRoche Inc.), cefpira- clearly established. Although the effects ofprotein binding in mide (Sumitomo Pharmaceutical Co., Ltd.), ceftazidime serum on penetration into extravascular fluid and tissues has (Glaxo Japan, Ltd.), and cefuzonam (Lederle Japan, Ltd.). been studied in various animal (5, 7, 12, 31) and human (2, 3, Bacteria. A standard strain of S. aureus Smith, a clinical 9, 27, 28, 33) models, the results were contradictory. Some isolate of K. pneumoniae 1, and K. pneumoniae B-54 were of these studies show that the penetrability is primarily used. These strains were stored at 4°C on heart infusion agar dependent on the free fraction in serum (2, 3, 9, 12, 27, 28, slants (Difco Laboratories, Detroit, Mich.). 33), whereas others show the contrary (5, 7, 31). The effect Antimicrobial susceptibility tests. MIC tests were carried of protein binding on therapeutic efficacy was also examined out by serial twofold dilution in Mueller-Hinton agar (Difco) in various animal models (1, 16, 17, 19, 20, 25), but the in petri dishes. The test organisms were incubated overnight results were inconclusive. The clinical significance of the in Mueller-Hinton broth (Difco). Overnight cultures were effect of protein binding on therapeutic efficacy is still diluted 100-fold with Mueller-Hinton broth, and 0.01 ml of an controversial. inoculum containing 104 CFU was applied to agar plates with In the study reported here, we investigated the effect of a multipoint replicating apparatus. These plates were incu- protein binding in serum on the therapeutic efficacy of eight bated at 37°C for 18 h. The MIC was the lowest antibiotic cephem antibiotics which markedly differ in their extent of concentration that inhibited macroscopic colonial growth binding and in their in vitro activities against the test strains. during this incubation. The therapeutic efficacy of each compound was evaluated Protective effect against intraperitoneal infection. Male against intraperitoneal infections with Staphylococcus au- Slc-ICR strain mice (age, 4 weeks; weight, 21 to 24 g) were reus or Klebsiella pneumoniae and pneumonia with K. used. S. aureus Smith and K. pneumoniae 1 were cultured pneumoniae, because the discrepancies of the results of overnight on Trypticase soy agar (BBL Microbiology Sys- studies on therapeutic efficacy may be due to differences in tem, Cockeysville, Md.) and heart infusion agar (Difco), animal models. The relationship among the therapeutic respectively, at 37°C. Each test strain was suspended in 5% activity, total or free antibiotic concentrations in serum, bacteriological mucin (Nacalai Tesque Inc., Kyoto, Japan), total lung homogenate levels, and in vitro susceptibility was and 0.5 ml (approximately one minimum lethal dose) was investigated to clarify the factors that affect therapeutic injected intraperitoneally into the mice. Eight mice were efficacy. allocated to each of four dosage levels, which were separated from one another by fivefold increments. The top dosage levels were as follows: ceftazidime, 50 mg/kg of body weight; ceftizoxime, 25 mg/kg; cefotiam, 5 mg/kg; cefmetazole, 25 mg/kg; cefpiramide, 5 mg/kg; cefazolin, 1 mg/kg; * Corresponding author. cefuzonam, 25 mg/kg; and ceftriaxone, 25 mg/kg for S. 17 18 TAWARA ET AL. ANTIMICROB. AGENTS CHEMOTHER.

aureus infection. For K. pneumoniae infection, the top percent contaminating blood in the supernatants. The level dosage levels were as follows: ceftazidime, 1 mg/kg; cefti- in lungs (CL; micrograms per gram) was calculated by the zoxime, 0.2 mg/kg; cefotiam, 5 mg/kg; cefmetazole, 25 following equation (11): CL = [CLH - CB (HbL/HbB)] x 3, mg/kg; cefpiramide, 50 mg/kg; cefazolin, 25 mg/kg; cefuzo- where CLH and CB are the antibiotic concentrations in the nam, 5 mg/kg; and ceftriaxone, 1 mg/kg. Appropriate doses supernatants of lung homogenate and blood, respectively, in 0.25 ml of distilled water were given subcutaneously 1 h and HbL and HbB are the hemoglobin concentrations in the after challenge. The mice were observed for 4 days. The supernatants of lung homogenate and blood, respectively. protective effect of the test drugs was expressed in terms of The total and free levels in serum were also determined by the 50% effective doses (ED50s; in milligrams per kilogram) the methods described above to investigate the relationship calculated by the probit method (4). between the level in lung and the total or free level in serum. Therapeutic effect against experimental pneumonia. Male Protein binding in serum. The percentage of protein bind- Slc-ICR strain mice (age, 4 weeks; weight, 25 to 28 g) were ing of the test antibiotics in serum was calculated from the used in groups of five each. Mice were anesthetized by data on the total and free levels in serum after a single intravenous injection of 1.5 mg of pentobarbital sodium. K. subcutaneous dose of 20 mg/kg. Each value for a drug pneumoniae B-54 that was cultured overnight at 37°C on represents the mean binding percent at each sampling time. Trypticase soy agar was suspended in physiological saline, Binding to lung homogenate. Lungs were removed from and 0.025 ml (5.0 x 105 CFU) was inoculated intranasally. freshly killed mice, and blood was rinsed away by washing it Doses of 0.4, 2, or 10 mg of ceftriaxone or ceftizoxime per kg in phosphate buffer. Lungs were blotted with filter paper and or 2 or 10 mg of cefuzonam, ceftazidime, cefotiam, or homogenized as 33% (wet weight) in phosphate buffer with a cefpiramide per kg were administered subcutaneously 4, 24, Polytron homogenizer. The binding to whole lung tissue and 36 h after bacterial inoculation. The mice were killed 48 was h after challenge, and the lungs were homogenized in phys- homogenate at an antibiotic concentration of 30 jig/ml iological saline. Serial 10-fold dilutions of homogenates in determined by the ultrafiltration method with Visking tubing. saline were prepared, and the numbers of viable organisms After centrifugation at 1,000 x g for 30 min, the concentra- in 1 ml of the diluted homogenates were measured by the tions in the ultrafiltrate were determined by the microbiolog- pour plate method. ical assay. The hemoglobin level in the lung homogenate was Antibiotic concentrations in sera and lungs. Male Slc-ICR determined by the method described above to estimate the strain mice (age, 4 weeks; weight, 25 to 28 g) were used in percentage of contaminating blood. Control samples con- groups of 10 each to determine levels in sera and in groups of taining 30 [.g of the test antibiotics per ml in phosphate buffer three each to determine levels in lungs. The antibiotics were containing the same volume of blood as the lung homogenate given subcutaneously in a single dose of 20 mg/kg for the were handled in the same way. The binding (percent bound determination of levels in serum and in a single dose of 50 [P]) was calculated by the following equation: P = 100 (1 - mg/kg for the determination of levels in lungs. X/Y), where X and Y are the concentrations of drug in the Blood specimens were taken from the hearts of the mice ultrafiltrate of the supernatant of lung homogenate and and serum was separated. The total levels of the test phosphate buffer, respectively. antibiotics in serum were determined by the disk plate Statistical analysis. It is known that there is a linear diffusion technique by using Bacillus subtilis ATCC 6633 as relationship between the probit value of the survival rate and the test organism for cefazolin, cefotiam, and cefuzonam and the logarithm of the dose (4). The in vivo therapeutic efficacy Escherichia coli ATCC 39188 for cefmetazole, ceftizoxime, is influenced by a number of factors. The important factors ceftriaxone, cefpiramide, and ceftazidime. Standard solu- affecting the protective effect against systemic infection are tions for bioassay were prepared with serum. in vitro antibacterial activity and levels of drug in serum. The concentrations of the free fraction in the serum were Assume that there are the following relationships between determined by the ultrafiltration method with a Visking levels in serum and the in vitro antibacterial activity of drug tubing (size, 8/32; Wako Chemicals, Ltd., Osaka, Japan). A and those of drug B: (i) for the area under the concentra- After centrifugation at 1,000 x g for 30 min, the antibiotic tion-time curve (AUC) of the level in serum, the AUC of concentrations in the ultrafiltrate were determined by the drug A = a, and the AUC of drug B = k. a; and (ii) for disk diffusion assay. Standard solutions for bioassay were antibacterial activity in vitro, the MIC of drug A = b, and the prepared with 67 mM phosphate buffer (pH 7.0). Control MIC of drug B = k2. b. At a certain dose the effective samples containing 32 or 4 ,ug of the test antibiotics per ml in amount of drug B is [AUC(B)/AUC(A)] x [MIC(A)/ phosphate buffer were handled in the same way to measure MIC(B)], which is k1/k2 times higher than that of drug A. the extent of the test antibiotics bound to the Visking tubing. Drug A must be given at a dose that is kl/k2 times the dose of The levels of the test antibiotics in lungs were determined drug B to yield the same protective effect as drug B. in mice uninfected and infected with K. pneumoniae B-54 24 Therefore, the survival rates of drug B at certain doses of d, h after challenge. The mice were killed at specified intervals and d2 are equal to those of drug A at doses of (kllk2) d, and after drug dosing. The lungs were removed and blotted with (kllk2) d2, respectively, as illustrated in Fig. 1. Conse- filter paper. The lungs of the mice in each group were pooled quently, the relationship of the ED50 between drugs A and B and homogenized with a Polytron homogenizer after the is shown by the following equation: ED50 of drug B = addition of 2 ml of phosphate buffer per g of lung. The (k2/kl)- ED50 of drug A. If the AUC/MIC is regarded as one homogenates were centrifuged at 10,000 x g for 10 min. The variable x and the ED50 as the other variable y, the relation- drug concentration in each supernatant was bioassayed by ship between x and y is shown by the following equations. using a standard curve prepared with the supernatant of lung For drug A, x = alb, and y = ED50 (drug A); for drug B, x = homogenates of mice uninfected or infected with K. pneu- (k1lk2) (alb), and y = (k2lkl) ED50 (drug A); and x y = moniae B-54. The hemoglobin levels in supernatants and (alb). ED50 (drug A). Since (alb). ED50 (drug A) is con- blood specimens were determined by the cyanmethemoglo- stant, l/y = k x (where k is constant). This equation shows bin method (15) by using Acuglobin (Ortho Diagnostic Sys- that there is a linear relationship between 1/ED50 and AUC/ tems Inc., Raritan, N.J.) as a standard to estimate the MIC. We examined whether this hypothesis is true. VOL. 36, 1992 PROTEIN BINDING AND THERAPEUTIC EFFICACY 19

This is illustrated in Fig. 2 (S. aureus) and Fig. 3 (K. pneumoniae), in which the 1/ED50s are plotted against AUC of total drug/MICs or AUC of free drug/MICs. In the systemic infection with S. aureus (Fig. 2), the correlation was poor when the 1/ED50s were plotted against the AUC of "4 total drug/MICs, especially in the case of ceftriaxone and 0 cefuzonam, which had high levels of protein binding. On the $4. other hand, a close correlation cold be seen when the /drug A 1/ED50s were plotted against the AUC of free drug/MICs (r = 0.97); this close correlation included ceftriaxone and cefuzonam. Similarly, the correlation between the 1/ED50s and the AUC of free drug/MICs was closer (r = 1.00) than $4. that between the 1/ED50s and the AUC of total drug/MICs in the systemic infection with K. pneumoniae (Fig. 3). These results show that protein binding in serum unfavorably affects the protective effect of the cephem antibiotics tested against systemic infection. d2 d, (k/k)d (kl/2 Effect of protein binding in serum on therapeutic efficacy Dose (log) against experimental pulmonary infection. The effect of protein binding in serum of six cephem antibiotics on their FIG. 1. on the Hypothesis relationship among protective effect, therapeutic efficacies was examined in experimental pneu- in vitro antibacterial activity, and level in serum of drugs A and B. monia induced by intranasal inoculation of K. pneumoniae B-54. The drugs were given subcutaneously 4, 24, and 36 h after bacterial inoculation. The relationship among the num- RESULTS ber of viable cells in the lungs 48 h after challenge, the MIC Effect of protein binding in serum on the protective effect for the test strain, and the levels of drug in lung tissue or against experimental systemic infection. The results of the serum after a single subcutaneous dose of 50 mg/kg was effect of protein binding in serum on the protective effect investigated. The results are summarized in Table 2. In the against experimental systemic infection are summarized in control group, the number of CFU in the lung increased to Table 1 for protein binding in mouse serum, the AUC of total approximately 1,000-fold 48 h after challenge, and all the and free drug after a single subcutaneous dose of 20 mg/kg, mice had died by 72 h. In the group treated with ceftriaxone, the in vitro activity (MIC), and the in vivo activity (ED50) the viable bacterial counts decreased to less than 10' CFU against intraperitoneal infection with S. aureus or K. pneu- per lung even at a dose of 0.4 mg/kg. Ceftizoxime at a dose moniae. In the control groups of these infection models, 103 of 2 or 10 mg/kg produced an almost completely therapeutic to 104 CFU of bacteria was recovered from blood, lung, response similar to that attained in mice given ceftriaxone, liver, and kidney 1 h after inoculation. The numbers of CFU but it was less effective than ceftriaxone at a dose of 0.4 in blood and these organs increased to 106 to 107 CFU 6 h mg/kg. The effects of cefuzonam and ceftazidime were after inoculation, and all the mice died within 48 h. The almost the same as those of ceftriaxone and ceftizoxime at protective effects of cephem antibiotics against intraperito- doses of 10 mg/kg, but they were weaker at doses of 2 mg/kg. neal infection varied markedly, with the ED50s ranging from Treatment with 10 mg of cefotiam or cefpiramide per kg 0.20 to 3.97 mg/kg for S. aureus and from 0.04 to 4.57 mg/kg resulted in a slight therapeutic effect, and no effect of either for K. pneumoniae. This variation in ED50s could not be drug was obtained at a dose of 2 mg/kg. This difference in the explained by the MICs or by the AUCs only. Moreover, therapeutic effects could not be explained by the MICs or by there was no direct correlation between the ED50s and the the AUCs of levels in lung only. The therapeutic effects did degree of protein binding. However, a close correlation was not directly correlate with the degree of protein binding. seen between the 1/ED50s and the AUC of free drug/MICs. However, there was a close correlation between the thera-

TABLE 1. Protein binding in serum, protective effect against intraperitoneal infection, and AUC of level in serum in mice for each of the cephem antibiotics tested AUC (1kg- h/ml) ofb: S. aureus K. pneumoniae Antibiotic Protein binding (%)a Total drug Free drug MIC (Lg/ml)c ED50 (mglkg)d MIC (>jg/ml)C ED50 (mg/kg)d Ceftazidime 27.2 ± 4.4 28.5 18.3 6.25 3.97 0.20 0.18 Ceftizoxime 29.5 ± 1.9 15.2 12.3 6.25 1.99 0.03 0.04 Cefotiam 48.3 ± 7.9 10.8 7.4 0.78 0.56 0.20 1.17 Cefmetazole 54.0 ± 0.8 14.1 7.5 1.56 2.35 1.56 2.88 Cefpiramide 71.3 ± 1.4 20.8 5.9 1.56 0.99 1.56 3.78 Cefazolin 80.8 + 1.1 38.9 8.0 0.39 0.20 1.56 4.57 Cefuzonam 93.5 ± 0.6 27.6 2.0 0.39 1.72 0.10 0.33 Ceftriaxone 94.2 ± 1.1 135.0 5.3 3.13 1.99 0.05 0.15 ± a Each value represents the mean standard deviation of the binding rate at each sampling time after a single subcutaneous dose of 20 mg/kg. b Drugs were given subcutaneously at a dose of 20 mg/kg. Agar dilution method. d Drugs were given subcutaneously 1 h after intraperitoneal challenge with test organisms. The ED50s were calculated by the probit method after observation for 4 days. 20 TAWARA ET AL. ANTIMICROB. AGENTS CHEMOTHER.

6 Total Free A

4. r-l

1-10' 0'at 0 Si u} - Q 04 y-0.26x 0.39 r.i 0 r=0.97

0 25 50 75 100 0 10 20 30 AUC of total serum level/MIC (h) AUC of free serum level/MIC (h) FIG. 2. Effect of protein binding in serum on the protective effect of cephem antibiotics against intraperitoneal infection with S. aureus Smith in mice. Values of 1/ED50 and AUC/MIC were calculated from the data in Table 1. Symbols: 0, ceftazidime; A, ceftizoxime; L, cefotiam; O, cefmetazole; 0, cefpiramide; A, cefazolin; U, cefuzonam; *, ceftriaxone. peutic effect and the AUC of the level in lung/MIC. A close in which the AUCs of levels in lung are plotted against those correlation was also seen between the therapeutic effect and of levels in serum. A close correlation was seen between the AUC of total level in serum/MIC but not between the total levels in serum and those in lung, but not between free therapeutic effect and the AUC of free level in serum/MIC. levels in serum and those in lung. The results indicate that The lung level AUCs of ceftriaxone and cefuzonam were the serum protein binding in serum does not directly affect higher than those of ceftizoxime, ceftazidime, and cefotiam the penetration into the lung under the experimental condi- in proportion to the total levels in serum. In addition, tions used in this study. The levels in lungs in mice infected ceftriaxone and cefuzonam produced the therapeutic effects intranasally with K. pneumoniae B-54 24 h after challenge expected from their MICs and levels in lung. These results were compared with those in normal mice. As shown in Fig. suggest that protein binding in serum does not directly affect 6, the lung level AUCs of cephem antibiotics with high levels penetration into the lungs and, consequently, did not unfa- of protein binding in infected mice were higher than those in vorably affect the therapeutic activities in the pneumonia normal mice. On the other hand, there was not a significant model used in this study. difference between lung level AUCs in infected mice and Effect of protein binding in serum on penetration into lung. those in normal mice for cephem antibiotics with low levels As shown in Table 2, the penetrations of ceftriaxone and of protein binding. The results suggest that protein binding in cefuzonam into the lung were good, despite their high degree serum also did not unfavorably affect the penetration of drug of protein binding in serum. To clarify whether protein into the lung in infected mice in this study. binding in serum affects the penetrations of the cephem Relationship between the binding to protein in serum and antibiotics tested into the lung, the relationship between lung homogenate. The binding of drug to whole lung tissue levels in serum and lung was investigated in mice uninfected homogenate of mice was determined at a concentration of 30 and infected with K. pneumoniae B-54. The results are ,ug/ml and was compared with that to protein in serum. The shown in Fig. 4 (uninfected mice) and Fig. 5 (infected mice), cephem antibiotics tested were 2.7 to 10.9% bound to lung

A Total Free

l l

0' I' -4 I -I 0'

0 0 '4 Y-0.058X + 0.16 0 r-1.00 .

100 200 300 0 200 400 600 AUC of total serum level/MIC (xlO-l) AUC of free serum level/MIC (h) (h) FIG. 3. Effect of protein binding in serum on the protective effect of cephem antibiotics against intraperitoneal infection with K. pneumoniae 1 in mice. Values of 1/ED50 and AUC/MIC were calculated from the data in Table 1. Symbols: 0, ceftazidime; A\, ceftizoxime; a, cefotiam; O, cefmetazole; 0, cefpiramide; A, cefazolin; *, cefuzonam; *, ceftriaxone. VOL. 36, 1992 PROTEIN BINDING AND THERAPEUTIC EFFICACY 21

TABLE 2. Protein binding in serum, therapeutic efficacy against experimental pneumoniae induced by K. pneumoniae, and AUC of levels in serum and lung of mice for each of the cephem antibiotics tested AUCf AUC/MIC ~~~~~~Bacterial tecovery Antibiotic' Protein binding Dose in the lung (loglo MgC Total drug Free drug in Lung Total drug Free drug (%)b (mgfkg)c CF ug' (li/mi)C in serum serum Lung (ag h/mI) (sag h/mi) (p.g. h/g) in serum in serum Ceftriaxone 94.2 ± 1.1 0.4 <1.14 2 <1.00 0.05 381.7 37.1 99.5 7,634 742 1,990 10 <1.00 Ceftizoxime 29.5 ± 1.9 0.4 5.56 ± 0.84 2 <1.31 0.03 53.6 44.3 17.5 1,787 1,477 583 10 <1.00 Cefuzonam 93.5 ± 0.6 2 2.35 ± 0.46 0.10 101.1 12.6 20.2 1,011 126 202 10 <1.76 Ceftazidime 27.2 ± 4.4 2 7.03 + 0.20 0.20 75.6 55.5 17.3 378 278 87 10 <1.00 Cefotiam 48.3 ± 7.9 2 8.06 ± 0.08 0.20 37.4 31.8 10.2 187 159 51 10 5.43 ± 0.73 Cefpiramide 71.3 ± 1.4 2 7.83 + 0.16 1.56 65.6 18.1 22.5 42 12 14 10 6.73 ± 0.39 Control 8.54 ± 0.12 a Drugs were given 4, 24, and 32 h after intranasal inoculation of 5 x iOs CFU of K. pneumoniae B-54. b Each value represents the mean + standard deviation of the binding rate at each sampling time after a single subcutaneous dose of 20 mg/kg. All doses were given three times. d Bacterial recovery in the lung 48 h after challenge is indicated (values are means ± standard deviations of five mice). Agar dilution method. f Drugs were given subcutaneously at a dose of 50 mg/kg 24 h after intranasal inoculation of 5 x 105 CFU of K. pneumoniae B-54.

homogenate. No correlation was seen between the binding not only the antibacterially active fraction of an antibiotic to protein in serum and that to lung homogenate. The results but also the penetrability into extravascular fluid. Despite suggest that almost all the drug molecules existed as the extensive studies about this subject, the effect of protein antibacterially active form in the lung. binding of antibiotics on therapeutic efficacy is not clearly established. This may not necessarily be surprising, because DISCUSSION therapeutic efficacy is affected by various factors, such as antibacterial and pharmacokinetic properties. There is con- It has been shown that protein binding has two important siderable agreement on the effect of protein binding on in effects from a chemotherapeutic point of view. It influences vitro antibacterial activity; that is, only the unbound fraction

80 ' Total Free

A 60 60' 1.1

.4 0' 0 -1 > 40. * 40 r.

.4 44 0 20- y-0.18x + 3.20 0 20. u u 0 r-0.99 0 a A 0

0 100 200 300 400 0 20 40 60 80 AUC of total serum level (pg/ml h) AUC of free serum level (pg/ml h)

FIG. 4. Effect of protein binding in serum on the penetration of cephem antibiotics into the lung in normal mice. Drugs were given subcutaneously at a dose of 50 mg/kg. Symbols: 0, ceftazidime; A, ceftizoxime; O, cefotiam; 0, cefpiramide; *, cefuzoname; *, ceftriaxone. 22 TAWARA ET AL. ANTIMICROB. AGENTS CHEMOTHER.

120. 1201 Total Free

A ts 90 tri 9O.- - N,

1-0'

@ 60. * 60

I4 H 0% :3 4 3 30.@ 30 * U A 0 0 0

100 200 300 400 20 40 60 AUC of total serum level (pg/ml h) AUC of free serum level ()pg/ml h) FIG. 5. Effect of protein binding in serum on the penetration of cephem antibiotics into the lungs of mice infected with K. pneumoniae B-54. Drugs were given subcutaneously at a dose of 50 mg/kg 24 h after intranasal inoculation of 5 x 105 CFU of K. pneumoniae B-54. Symbols: 0, ceftazidime; A, ceftizoxime; O, cefotiam; 0, cefpiramide; M, cefuzonam; *, ceftriaxone. is active against bacteria (23, 29). However, results of demonstrated that extravascular penetration is affected by studies on the effects of protein binding on penetration into the degree of protein binding in the extravascular fluid. The extravascular fluid and tissues are contradictory. These effect of administration method on the penetrability has been discrepancies may be explained by differences in (i) experi- shown in studies in which animal (8, 30) and human (14) mental model and severity of inflammation and (ii) method of models were used. Therefore, the effect of protein binding in drug administration. For example, Peterson et al. (21, 22) serum on therapeutic efficacy may depend on the infection model and the method of treatment. For these reasons, we studied the therapeutic efficacy of eight cephem antibiotics 120. with different degrees of protein binding using two infection models and investigated the factors that affected the therapeutic results. In the systemic infections induced by S. aureus or K. pneumoniae, therapeutically effective antibiotic doses correlated more closely with the levels of the free antibiotic in 90~ serum than with those of total antibiotic. The results indicate that the protective effect against systemic infection is determined mostly by the MIC and free drug levels. Thus, our g results demonstrate an inhibitory effect of protein binding in .4 serum on the in vivo efficacy of a drug against systemic

"I infections. On the other the effect of in serum 60' hand, protein binding on therapeutic efficacy against localized infection differed from that against systemic infection. Mattie et al. (16) investigated this subject in renal infection, but that model did "4 not seem to be appropriate for this purpose. Drug concen- 0 trations in the kidney are not determined only by the 03 penetration from the blood because it is a drug-excreting organ. For this reason, we chose pneumonia as a localized infection model. In the experimental pneumonia induced by intranasal inoculation of K. pneumoniae, the direct inhibitory effect ofprotein binding in serum on therapeutic efficacy

I was not seen. The cephem antibiotics with high levels of WI~~~~~~~~~~~~~~~~ protein binding in serum, such as ceftriaxone and cefuzonam, penetrated into the lung in proportion to their total ceftriaxone cefpiramide cefuzona ceftazidime ceftizoxime cefotiam levels in serum, but not to their free levels in serum. The FIG. 6. AUCs of lung levels of cephem antibiotics in mice antibiotics with high levels of protein binding showed higher uninfected (hatched bars) and infected with K. pneumoniae B-54 levels in the lungs of mice with pneumonia than in those of (stippled bars). Drugs were given subcutaneously at a dose of 50 normal mice. This can be explained by protein leakage from mg/kg 24 h after intranasal inoculation of 5 x 105 CFU of K. blood vessels because of inflammation caused by the infec- pneumoniae B-54. tion. The cephem antibiotics tested in this study were bound VOL. 36, 1992 PROTEIN BINDING AND THERAPEUTIC EFFICACY 23 only slightly to whole lung tissue homogenates of mice, and administration method. Waterman et al. (31) showed that there was no correlation between the binding to protein in total concentrations of cefazolin and cephaloridine in inter- serum and that to lung homogenate. The results are in stitial fluid correlated with the total levels in serum but not agreement with the data of Kunin (13), who demonstrated with' the free levels in serum in dogs after a single intrave- that the activity of P-lactam, antibiotics is not inhibited by nous administration. Levels of ceftriaxone and cefuzonam in 25% homogenates of various rabbit tissues. Therefore, it the lungs of rats after a single dose (6, 10) were not lower seems that almost all the drugs in the lung exist as the than those of ceftizoxime (18). On the other hand, Shimizu et antibacterially active fraction. al. (26) showed that tissue cage fluid levels of 3-lactam The results presented above show that there is a different antibiotics given in a continuous infusion correlated with the relationship between the outcome of treatment of two infec- free levels in the serum of rabbits. This different effect of tion models and antibiotic concentration in serum. Thera- protein binding on extravascular penetration can be ex- peutic efficacy against the intraperitoneal infection model plained as follows, according to the hypothesis of Rolinson correlated with free levels of drug in serum. In this model, (24). The effect of protein binding on antibiotic levels in the the numbers of bacteria recovered from blood, lung, liver, tissues depends on the half-life in serum. When drugs are and kidney 1 h after inoculation were not significantly given in a drip infusion, levels in serum are maintained for a different. This result suggests that intraperitoneally inocu- long period of time and the levels in tissue will eventually lated bacteria entered the blood vessels and then infected equilibrate with the free levels in serum. On the other hand, these organs. If inoculated bacteria colonized a target organ equilibrium will not be reached after a subcutaneous injec- at first and then infected other organs via blood vessels, a tion because of a limited period of time for diffusion to take time lag of bacterial recovery would be observed in the place.' Thus, protein binding did not directly affect the secondarily infected organs. It seems that killing of bacteria therapeutic efficacy on experimental pneumonia in this in blood- is important for survival of mice in'this model. study. Therefore, therapeutic efficacy correlated with free levels of The following factors are regarded as artifacts that may drug in serum because protein-bound drug is not antibacte- modify our conclusions. (i) There may have been differences rially active. Our results are consistent with the findings of in the dosage regimen, i.e., a single dose 1 h after challenge Merrikin et al. (17), who showed the direct correlation for the intraperitoneal infection model and three doses at 4, between the in vivo activity against intraperitoneal infection 24, and 36 h after challenge for the pneumonia model. with S. aureus and protein' binding in serum in mice, using Therapeutic efficacy against the pneumonia model after a isoxazolyl penicillins with similar in vitro activities and single dose was investigated. Drugs were given subcutane- pharmacokinetic properties. This phenomenon was also ously 4 h after bacterial inoculation, and the numbers of demonstrated by Muckter et al. (19). Although we used the viable bacteria in the lungs were determined 8 h after AUC as the pharmacokinetic parameter in Fig. 2 and 3, we challenge. The differences in the therapeutic effects of each also> analyzed the correlation by means of peak level in of the drugs after three doses was more remarkable than that serum and half-life in serum as pharmacokinetic parameters after a single dose, but the relationship between the thera- instead of AUC, but the correlation was poorer (data not peutic effects did not significantly change (data not shown). shown). Furthernmore, we examined the correlation between (ii) There may have been drug accumulation after the three the therapeutic efficacy and the time when the concentration repeated doses (at 4, 24, and 36 h after challenge) in the in serum exceeded the MIC for the test strains, but the pneumonia model. The elimination half-life for ceftriaxone correlation was also poorer (data not shown). from serum was 1.4 h in mice after a single' subcutaneous On the other hand, therapeutic efficacy against pneumonia administration. The half-lives of the other antibiotics were correlated with total levels-in serum. It is a matter of course 0.2 to 0.38 h. Therefore, it seems that the drugs did not that the antibiotic concentration in the lung tissue governs accumulate in the lungs under the experimental conditions of the outcome of treatment of pneumonia. It is accepted that this study. (iii) There may have been an effect of the only the free fraction can diffuse out from the vascular inoculum size on the MIC. The relationship between thera- compartment into the tissues. Therapeutic efficacy against peutic efficacy and the AUC/MIC was investigated by using pneumonia should correlate with free levels in serum accord- MICs at an inoculum size of 106 CFU. The MICs increased ing to this theory. Sande et al. (25) showed that the penetra- approximately twofold, but' the relationship between the tion of cephalosporins into the cerebrospinal fluid of rabbits MICs of cach of the drugs was not significantly different infected with Streptococcus pneumoniae varied inversely from that at an inoculum size of 104 CFU (data not shown). with binding to protein in serum, but the therapeutic efficacy Therefore, the effect of inoculum size on MIC did not modify was similar, although the MICs and the bactericidal effects in our conclusions. cerebrospinal fluid were similar. Our results were consistent Our findings indicate that the effect of protein binding in with their findings, in that therapeutic efficacy was not serum on therapeutic efficacy is influenced by the infection directly related to protein binding in serum, but they were model usedi. There is no clinical evidence that a highly bound contradictory in that extravascular penetration was unaf- druig is inferior to a less bound drug, as pointed out by Wise fected by protein binding in our study. In rats with pneumo- (32). Therefore, the experimental infection model and the nia caused by K. pneumoniae, Bakker-Woudenberg et al. (1) mode of drug administration should be carefully designed to demonstrated that protein binding in serum unfavorably estimate the clinical efficacies of highly protein-bound drugs. affected both'the penetration into lungs and the therapeutic ACKNOWLEDGMENTS efficacy. The discrepancies may be explained by the differences in the duration of drug administration. Drugs were We thank Lance R. Peterson for kind help and advice and given subcutaneously 4, 24, and 32 h after challenge in this Toshiharu Hirose, Yoshimi Matsumoto, and Shoji Nakamoto for discussions. study, but Sande et al. (25) gave drugs in a continuous 8-h criticism and infusion and Bakker-Woundenberg et al. (1) gave drugs in a REFERENCES continuous 65-h infusion. The effect of protein binding on 1. Bakker-Woudenberg, I. A. J. M., J. C. Van Den Berg, T. B. extravascular penetration of antibiotics depends on the Vree, A. M. Baars, and M. F. Michel. 1985. Relevance of serum 24 TAWARA ET AL. ANTIMICROB. AGENTS CHEMOTHER.

protein binding of cefoxitin and cefazolin to their activities ceftizoxime in animals after parenteral dosing. Antimicrob. against Klebsiella pneumoniae pneumonia in rats. Antimicrob. Agents Chemother. 17:157-164. Agents Chemother. 28:654-659. 19. Muckter, H., H. Sous, G. Poszich, and P. Arend. 1976. Compar- 2. Bergan, T., A. Engeset, and W. Olszewski. 1987. Does serum ative studies of the activity of ciclacillin and dicloxacillin. protein binding inhibit tissue penetration of antibiotics? Rev. Chemotherapy (Basel) 22:183-189. Infect. Dis. 9:713-718. 20. Nishida, M., and T. Murakawa. 1977. Exudate levels and 3. Bergan, T., A. Engeset, W. Olszewski, N. 0stby, and R. Solberg. bactericidal activity of cefazolin in a new local infection system 1986. Extravascular penetration of highly protein-bound flu- using rat granuloma pouches. Antimicrob. Agents Chemother. cloxacillin. Antimicrob. Agents Chemother. 30:729-732. 11:1042-1048. 4. Bliss, C. I. 1952. The statistics of bioassay. Academic Press, 21. Peterson, L. R., and D. N. Gerding. 1978. Prediction of cefazolin Inc., New York. penetration into high- and low-protein-containing extravascular 5. Carbon, C., A. Contrepois, N. Brion, and S. Lamotte-Barrillon. fluid: new method for performing simultaneous studies. Antimi- 1977. Penetration of cefazolin, cephaloridine, and cefamandole crob. Agents Chemother. 14:533-538. into interstitial fluid in rabbits. Antimicrob. Agents Chemother. 22. Peterson, L. R., L. L. Van Etta, C. E. Fasching, and D. N. 11:594-598. Gerding. 1984. Effect of protein binding on simulated intravas- 6. Fukazawa, H., H. Tahara, S. Ichihara, H. Tomisawa, and M. cular and extravascular kinetics of cefotaxime in an in vitro Tateishi. 1984. Disposition of ceftriaxone (Ro 13-9904) in rats model. Antimicrob. Agents Chemother. 25:58-61. (I): blood level profiles, tissue distribution and excretion of 23. Quinn, E. L., J. M. Colville, L. Ballard, D. Jones, and F. 14C-ceftriaxone. Chemotherapy (Japan) 32(Suppl. 7):136-147. Debman. 1962. Ampicillin antimicrobial activity and pharmaco- 7. Gerding, D. N., L. R. Peterson, D. C. Legler, W. H. Hall, and logical behavior with reference to certain gram-positive cocci, E. A. Schierl. 1978. Ascitic fluid cephalosporin concentrations: p. 339-349. Antimicrob. Agents Chemother. 1961. influence of protein binding and serum pharmacokinetics. Anti- microb. Agents Chemother. 14:234-239. 24. Rolinson, G. N. 1980. The significance of protein binding of 8. Gerding, D. N., L. L. Van Etta, and L. R. Peterson. 1982. Role antibiotics in antibacterial chemotherapy. J. Antimicrob. Che- of serum protein binding and multiple antibiotic doses in the mother. 6:311-317. extravascular distribution of ceftizoxime and cefotaxime. Anti- 25. Sande, M. A., R. J. Sherertz, 0. Zak, and L. J. Strausbaugh. microb. Agents Chemother. 22:844-847. 1978. Cephalosporin antibiotics in therapy of experimental 9. Hoffstedt, B., and M. Walder. 1981. Influence of serum protein Streptococcus pneumoniae and Haemophilus influenzae menin- binding and mode of administration on penetration of five gitis in rabbits. J. Infect. Dis. 137:s161-s168. cephalosporins into subcutaneous tissue fluid in humans. Anti- 26. Shimizu, K., T. Kumada, K. Totsuka, J. Katahira, T. Fukuda, microb. Agents Chemother. 20:783-786. H. Hasegawa, K. Kikuchi, K. Kumano, T. Hayashi, and Y. 10. Inokawa, Y., N. Yamashita, and H. Takei. 1986. Pharmacoki- Watanabe. 1988. Penetration of antibiotics into inflamation foci. netics of L-105 in experimental animals. Chemotherapy (Japan) Chemotherapy (Japan) 36:632-639. 34(Suppl. 3):105-118. 27. Shyu, W. C., R. Quintiliani, C. H. Nightingale, and M. N. 11. Klesel, N., M. Limbert, K. Seeger, G. Seibert, I. Winkler, and E. Dudley. 1988. Effect of protein binding on drug penetration into Schrinner. 1984. Cefodizime, an aminothiazolylcephalosporin. blister fluid. Antimicrob. Agents Chemother. 32:128-130. II. Comparative studies on the pharmacokinetic behavior in 28. Tan, J. S., and S. J. Salstrom. 1977. Levels of carbenicillin, laboratory animals. J. Antibiot. 37:901-909. ticarcillin, cephalothin, cefazolin, cefamandole, gentamicin, to- 12. Kunin, C. M. 1965. Effect of serum binding on the distribution bramycin, and amikacin in human serum and interstitial fluid. of penicillins in the rabbit. J. Lab. Clin. Med. 65:406-415. Antimicrob. Agents Chemother. 11:698-700. 13. Kunin, C. M. 1970. Binding of antibiotics to tissue homoge- 29. Tompsett, R., S. Schultz, and W. McDermott. 1947. Relation of nates. J. Infect. Dis. 121:55-64. protein binding to the pharmacology and antibacterial activity of 14. LeBel, M., S. Gregoire, M. Caron, and M. G. Bergeron. 1985. penicillins X, G, dihydro-F, and K. J. Bacteriol. 53:581-595. Difference in blister fluid penetration after single and multiple 30. Van Etta, L. L., G. R. Kravitz, T. E. Russ, C. E. Fasching, D. N. doses of ceftriaxone. Antimicrob. Agents Chemother. 28:123- Gerding, and L. R. Peterson. 1982. Effect of method of admin- 127. istration on extravascular penetration of four antibiotics. Anti- 15. Matsubara, T., and H. Okuzono. 1976. A new reagent for the microb. Agents Chemother. 21:873-880. HiCN method, abstr., p. 368. 16th International Congress of 31. Waterman, N. G., M. J. Raff, L. Scharfenberger, and P. A. Hematology 1976. Barnwell. 1976. Protein binding and concentrations of cephalo- 16. Mattie, H., W. R. 0. Goslings, and E. L. Noach. 1973. Cloxa- ridine and cefazolin in serum and interstitial fluid of dogs. J. cillin and nafcillin: serum binding and its relationship to anti- Infect. Dis. 133:642-647. bacterial effect in mice. J. Infect. Dis. 128:170-177. 32. Wise, R. 1983. Protein binding of P-lactams: the effects on 17. Merrikin, D. J., J. Briant, and G. N. Rolinson. 1983. Effect of activity and pharmacology, particularly tissue penetration. II. J. protein binding on antibiotic activity in vivo. J. Antimicrob. Antimicrob. Chemother. 12:105-118. Chemother. 11:233-238. 33. Wise, R., A. P. Gillet, B. Cadge, S. R. Durham, and S. Baker. 18. Murakawa, T., H. Sakamoto, S. Fukada, S. Nakamoto, T. 1980. The influence of protein binding upon tissue fluid levels of Hirose, N. Itoh, and M. Nishida. 1980. Pharmacokinetics of six P-lactam antibiotics. J. Infect. Dis. 142:77-82.