Pseudomonas Infections

Transcription

1 ANTrmcROa AGNm AND CHmEMOThRAPY, Dec. 1974, p American Society for Microbiology Vol. 6, No. 6 Printed in U.SA. Interaction of Purulent Material with Antibiotics Used to Treat Pseudomonas Infections RICHARD E. BRYANTI AND DIANNE HAMMOND Department of Medicine, Vanderbilt University Medical School, Nashville, Tennessee 37203, and the Veterans Administration Hospital, Nashville, Tennessee Received for publication 16 July 1974 To define factors contributing to the adverse prognosis of patients with gram-negative bacillemia and abscess formation, we studied the interaction between polymyxin B, colistin sulfate, gentamicin, or carbenicillin with purulent material. Carbenicillin activity was not significantly altered by incubation with pus. Equal volumes of antibiotic and purulent sediment decreased the effective concentration of polymyxin B, colistin sulfate, or gentamicin from 100 yg/ml to 3 to 6 ug/ml. One milliliter of purulent sediment bound more than 700,ug of gentamicin and 1,500,ug of polymyxin B or colistin sulfate. This effect occurred rapidly, proceeded at 4 and 37 C, was stable for 24 to 48 h, and was altered, but not abolished, by varying the ph of the solution. Antibiotic activity could be removed from pus by high concentrations of protamine sulfate, heparin, sodium chloride, or potassium chloride, suggesting binding rather than inactivation. The need for surgical drainage of abscesses is well established. However, in rare instances it may be necessary to rely exclusively on antibiotic therapy. Although it has been shown that adequate concentrations of penicillin are present within abscesses, little information is available regarding activity of other antibiotics within abscesses (4, 14). There are several reasons for examining problems of abscesses caused by gram-negative bacilli. Patients with such infections have a poor prognosis (2, 7). Abscesses may be multiple and inaccessible to drainage, especially in patients with Pseudomonas infection (10, 13). Although increased mortality of Pseudomonas infection may be explained in part by the frequency of infections in the elderly and infirm, it is likely that the proclivity of Pseudomonas to produce widespread infection with focal necrosis- may contribute significantly to its virulence (13). In addition, there is evidence to suggest that antibiotic activity in tissues might be impaired. Studies by Kunin demonstrated that aminoglycoside and polymyxin antibiotics were avidly bound by tissue homogenates of animals (6). Thus, purulent material in tissue could bind certain antibiotics and perhaps alter their efficacy in management of infection caused by gram-negative bacilli. Extensive experience in management of 'Present address: University of Oregon Medical School, Division of Infectious Disease, 3181 S.W. Sam Jackson Park Rd., Portland, Ore Pseudomonas infections has been obtained with only four currently available antibiotics; carbenicillin, colistin, polymyxin B, and gentamicin. The present studies evaluated activity of these agents after incubation with purulent material in order to gain insight applicable to management of patients with abscesses caused by Pseudomonas aeruginosa. MATERIALS AND METHODS Antibiotics and other reagents. Colistin sulfate was provided by Warner-Lambert Research Institute, U.S.P. Sodium heparin was obtained from Connaught Laboratories, Toronto. Other antibiotics and reagents were obtained as the injectable form: gentamicin (Garamycin; Schering) and protamine sulfate (Upjohn). All drugs were diluted with sterile distilled water. Purulent material. Pus used in these studies was collected from four patients with pneumococcal empyema. Antibiotic-free pus was sedimented by centrifugation for 10 min at 12,000 x g at 4 C in a refrigerated centrifuge. Sediment and liquid portions were separated. The liquid portion representing onefourth to one-fifth the volume of the whole pus was found to have no effect on antibiotic activity. Sedi-' ment was homogenized (Polytron-Kinematica, Lucern) briefly to produce a uniform mixture which was divided into small portions and frozen until used. As needed,' the sediment was thawed, mixed with a Vortex mixer, and measured with a disposable 1-ml syringe (without a needle). Binding characteristics of freshly prepared sedimented material were not significantly changed by homogenization or freezing. Homogenized tissue (0.3 ml) equal to approximately

2 VOL. 6, 1974 mg (wet weight) was used in all binding experiments unless otherwise noted. Antibiotic, at the concentration to be tested, was added to an equal volume of purulent sediment. The sample was then mixed with a Vortex mixer and incubated for 1 h at 37 C prior to centrifugation at 12,000 x g for 10 min. The antimicrobial activity of the supernatant was quantitated using the plate assay method described below. Each specimen was tested for its capacity to reduce antimicrobial activity. Reduction in activity was comparable with each specimen of pus from all four patients and corresponded quantitatively to values shown in Table 1. Studies examining the effect of ph and reversibility of drug binding were performed with purulent material from a single patient. Purulent material was examined microscopically to determine whether the processes of collection, homogenization, or storage caused overt change in appearance of pus sediment. Microscopic appearance of purulent material was not altered by manipulation involved in this study. Pus used in binding studies was not subjected to sonication. With the exception of studies examining the effect of ph on binding, pus was not exposed to extremes of ph. Preliminary studies demonstrated that there was less than a 10% variation in binding properties of freshly collected, unfrozen purulent sediment, of refrigerated abscess sediment, and of intentionally sonicated abscess sediment. Plate assay. Three strains of bacteria were used for the pour-plate assay studies. A clinical strain of Staphylococcus epidermidis was used for carbenicillin assay, Staphylococcus epidermidis ATCC 110 was used for gentamicin assay, and a clinical isolate of Klebsiella pneumoniae was used for colistin sulfate and polymyxin B assay. Bacteria were incubated ovemight in Penassay broth (Difco) prior to plating. Plates were made by mixing either 0.3 ml of an ovemight culture of S. epidermidis or 0.05 ml of an overnight culture of K. pneumoniae with 30 ml of melted (46 C) Penassay seed agar (Difco) and promptly pouring this mixture into a 150-mm petri dish. Wells of approximately 10-lambda volume were punched in the agar and removed with a sterile Pasteur pipette attached to a vacuum line. An Oxford automatic 10-lambda pipette was used to fill the wells with antibiotic standards and test samples. All samples and standards were run in duplicate. Samples of water or other solutions, i.e., NaCl, heparin, KCI, incubated with pus did not cause zones of inhibition on bacterial plates at concentrations tested. Antibiotic standards were diluted in water and plated in doubling dilutions from 200 Ag/ml to 1.56 ug/ml. Plates containing S. epidermidis were incubated at 37 C overnight, and plates containing Klebsiella strains were incubated at room temperature overnight. Zone sizes were measured in centimeters after projection of the plate image on a screen With a 3 M Five-Twenty-One overhead projector. Antibiotic activity remaining in the supernatant after incubation with purulent material was expressed in micrograms per milliliter. Since carbenicillin activity was not reduced by incubation with purulent material, the INTERACTION OF PUS WITH ANTIBIOTICS 703 fluid present in the purulent material was considered to be negligible. Effect of time and temperature. These studies were performed according to the format described above, with the exception that conditions of incubation were modified. The concentration of antibiotics shown in Table 2 was incubated with equal volumes of purulent material for either 5 min at 4 C, 5 min at 37 C, 60 min at 37 C, or 24 or 48 h at 37 C prior to centrifugation and determination of the antimicrobial activity of the supernatant. Effect of ph. The,effect of hydrogen ion concentration on antibiotic binding was determined by adjusting the ph of purulent material with buffered solutions of physiological sahne. Tris(hydroxymethyl)- aminomethane buffer (0.02 M) adjusted to ph 8.0 was used to obtain an alkaline ph, and 0.02 M N-tris(hydroxymethyl)methyl-2-aminoethane sulfonic acid buffer was used to reduce the ph of purulent material to 4.0. Equal volumes of phadjusted antibiotic solution and purulent material were incubated for 1 h at 37 C before specimens were centrifuged at 12,000 x g and the supernatant was removed. The ph of the supernatant was adjusted to ph 7.0 prior to determination of antibiotic activity. Removal of bound antibiotics. These studies were performed to determine factors which affected dissociation of antibiotics from purulent material. Purulent exudates which had been rotated with an equal volume of antibiotic for 1 h at 37 C were centrifuged at 12,000 x g, and the supernatant fluid was removed. Specimens of purulent material were then mixed with equal volumes of water, aqueous solution containing 1,000 U of heparin per ml or 4 mg of protamine sulfate per ml, or a 1 M solution of KCI or NaCl. Specimens were rotated for 1 h on a rotary drum at 37 C prior to centrifugation at 12,000 x g and determination of the antibiotic activity present in the supematant fluid. Standard antibiotic solutions were adjusted to contain the same concentration of heparin, protamine, KCI, or NaCl as the test specimens. Heparin, protamine, KCI, or NaCl solutions incubated with antibiotic-free pus did not inhibit bacteria in pour plates. In several tests, a second or third incubation and centrifugation period was used to determine the effect of repeated washing on removal of antibiotics from purulent material. Similarly, several studies used repeated washing with molar NaCl and 1,000 gg of heparin per ml after incubating antibiotic-sediment mixtures for 24 h in an attempt to remove bound antibiotics. Varied concentration of purulent sediment. These studies were performed according to the general method of incubation, except that the volume of purulent sediment was modified by appropriate dilution in water before addition to solutions containing antibiotics. Purulent sediment mixtures were diluted to achieve a final concentration by volume of 50, 40, 20, 10, 5, or 2.5% during incubation with antibiotic solutions. After 1 h of incubation, specimens were centrifuged and the antibiotic activity in the supernatant was determined. Efect of small quantities of purulent sediment.

3 704 BRYANT AND HAMMOND Equal volumes of solutions containing antibiotics were mixed with solutions of purulent sediment diluted 1:10 with water. The final concentration of antibiotics during incubation was 200, 100, 50, 25, 12.5, or 6,25 ug/ml, and the final concentration of purulent sediment was 5% by volume. Specimens were incubated 1 h at 37 C before centrifugation and determination of antibiotic activity in the supernatant. Because of the errors inherent in biological assay systems, the small volume of purulent sediment in this study was considered to have a negligible effect on calculation of antibiotic activity. RESULTS Incubation of antibiotics with purulent material. The effect of incubating antibiotics with pus is shown in Table 1. There is marked reduction in gentamicin, colistin, and polymyxin B levels after incubation with an equal volume of purulent material for 1 h. This was observed even when high concentrations of antibiotics were used. The accuracy of quantitating reduced activity was optimal when free drug concentrations of 4 to 16 jig/ml were observed, because those values were within the sensitive part of the antimicrobial assay. Purulent material appeared to have the greatest avidity for polymyxin B, because increasing concentrations of that antibiotic caused the least change in drug activity remaining after incubation. At lower concentrations, reduction in drug activity was comparable with colistin, polymyxin B, and gentamicin. Carbenicillin activity was not changed by incubation in pus. Effect of time and temperature. Changes in gentamicin, polymyxin B, and colistin levrels after incubation with purulent material appeared to represent binding of those agents by purulent material. To exclude the possibility of TABLE 1. Effect of incubating antibiotic solutions with sediment obtained from purulent material Antimicrobial activity of supernatant Concn of after incubation for 1 h with purulent drug being sediment (gg/ml)d incubated (pg/ml) Genta- Colistin Poly- Carbenimicin sulfate myxin B cillin 25 < < < , avalues represent mean of two studies performed in duplicate. Standard deviation of carbenicillin assay at 100-tsg concentration 4 24 gg; SE L 3.0 ;g. Standard deviation of gentamicin assay at 3.1- jg concentration Ag; SE = jpg. antibiotic destruction by purulent material, the effect of time, temperature, and ph on this reaction was tested. Changes in antibiotic activity after incubation with pus were not altered significantly by changes in temperature (Table 2). Reduction in activity proceeded as well at 4 C as it did at 37 C. In addition, it was noted that reduction in drug activity occurred rapidly and was virtually complete by 5 min. Incubation of purulent material with antibiotics for 24- or 48-h periods did not significantly alter the gentamicin or colistin activity remaining in the supernatant fraction. The finding that prolonged incubation did not change supematant activity suggested that equilibrium had occurred and that the interaction of pus with antibiotics was not primarily dependent on enzymatic destruction of the drug. Effect of ph. Table 3 shows the effect of ph on antibiotic interaction with pus. Reduction of gentamicin activity appeared to occur maximally at ph 7.0 and was inhibited at ph 4.0 or 8.0. However, the interaction of purulent material with polymyxin B or colistin was not altered significantly at an alkaline ph. Reduction of polymyxin and colistin activity after incubation with pus was less marked at ph 4.0. Removal of bound antibiotics. Table 4 demonstrates factors affecting removal of bound antibiotics from purulent material. Sterile distilled water removed very little antibiotic activity from pus which had been previously incubated with antibiotics. However, addition of high concentrations of either protamine sulfate or heparin, or molar solutions of NaCl or KCl, to purulent material previously incubated with TABLE 2. Effect of incubation time and temperature on antibiotic activity of solutions incubated with purulent sediment Time Temp Temp Antibiotic AN"MICROB. AG. CHIZMOTHER. Antibiotic activity. of solutions ~~~~~~(pg/ml). (C) ~~~~~Before After incuba- tion incubation 5 min 4 Gentamicin min 37 Gentamicin h 37 Gentamicin EO h 37 Gentamicin h 37 Gentamicin min 4 Colistin min 37 Colistin h 37 Colistin h 37 Colistin h 37 Colistin a Mean values of two studies, performed in duplicate.

4 VOL. 6, 1974 TABLE 3. Effect of ph on antibiotic activity of solutions incubated with purulent sediment ph Antibiotic Antibiotic activity of solutions (pug/ml)a Before After incubation incubation 4 Gentamicin Gentamicin Gentamicin 400 E Colistin Colistin Colistin Polymyxin B Polymyxin B Polymyxin B a All assays performed at ph 7.0. Mean values of two studies performed in duplicate. TABLE 4. Removal of antibiotics bound to purulent material Antibiotic activity (pg)a Antibiotic Wash fluid Present Present in in first purulent wash material solution Gentamicin Water (ph 6.4) 100 <2 Protamine Sulfate (4 mg/ml) Heparin (1, U/ml) 1 M NaCl 100 '223 1MKCl Colistin Water (ph 6.4) Protamine Sulfate (4 mg/ml) Heparin (1, U/ml) 1 M NaCl MKCI Polymyxin B Water (ph 6.4) Protamine sulfate (4 mg/ml) Heparin (1, U/ml) 1MNaCl M KCl amean values of two studies performed in duplicate. antibiotics increased antibiotic activity in the supematant. Roughly 10 to 15% of bound antibiotic was released with a single period oi incubation. A second washing procedure removed similar quantities of the drug. INTERACTION OF PUS WITH ANTIBIOTICS 705 Increasing the volume of the second wash solution of 1 M NaCl to 3.0 ml permitted recovery of 38,ug of gentamicin, 13,ug of polymyxin B, or 51 gg of colistin per ml, respectively. Repeated washing of purulent sediment containing 400 Mg of gentamicin or colistin with solutions containing 'both 1 M NaCl and 1,000,ug of heparin per ml permitted recovery of 375 Mg of gentamicin or 315 Mg of colistin, respectively. Bound antibiotics could not be completely removed from purulent material despite measures as extreme as sonication, acid extraction, or boiling. However, recovery was not significantly reduced when incubation preceded extraction by many hours. Even after incubation of 400 Mg of gentamicin with pus for 24 h, it was still possible to recover 344 gg of gentamicin. Varied concentration of purulent sediment. The effect of exposing a fixed concentration of antibiotic to different concentrations of purulent sediment is shown in Table 5. There was marked reduction of antibiotic activity after incubation with all concentrations of purulent material. This occurred despite greater dilution of antibiotics in mixtures containing smaller concentrations of purulent material. Effect of small quantities of purulent sediment. The effect of exposing varied concentrations of antibiotics to 5% (by volume) of purulent material is shown in Table 6. Reduction of drug activity occurred with all concentrations of antibiotics except the 200-,ug/ml concentration of gentamicin. Persistence of drug activity was less marked when antibiotic concentrations were 50 or 100,g/ml. This suggested that saturation of binding sites in the purulent material had occurred. DISCUSSION Incubation of gentamicin, polymyxin B, or colistin sulfate with sediment obtained from pus causes a dramatic reduction in supernatant TABLE 5. Effect of incubating various quantities of purulent sediment with antibiotics Antimicrobial activity after Concn of purulent incubation of 50 pg of antibiotics per sediment ml with purulent sediment for 1 h (volume %) (pg/ml) Gentamicin Colistin Polymyxin

5 706 BRYANT AND HAMMOND TABLE 6. Effect of incubating antibiotic solutions with 5% final volume of purulent material Antibiotic Antbioic concn Antimicrobial activity remaining after incubation for 1 h with 5% final volume of purulent material (Aggml) Gentamicin Colistin Polymyxin <1 < <1 <1 < <1 <1 <3 antibiotic activity. This reduction occurs rapidly, is not altered by prolonged incubation or low temperature, and is demonstrable with high concentrations of antibiotics and low concentrations of purulent sediment. The reaction is altered by acid ph and appears to be reversible at least in part by addition of increased concentrations of heparin or protamine sulfate or a 1 M solution of NaCl or KCI. These findings are best explained by the consideration that the particulate components of pus are capable of binding gentamicin, polymyxin B, or colistin sulfate. The liquid component of pus did not alter gentamicin, polymyxin B, or colistin activity. Thus, studies could have been performed in whole pus. This was not done in part because of the greater experience with drug stability in water and in part because of the greater ease in performing antibiotic assay studies and calculating volume of distribution of antibiotics when the solid component of pus was used. A number of in vitro studies have suggested that antimicrobial effectiveness is dramatically altered by changes in milieu. Antibiotic activity may be suppressed by changes in ph, ionic strength, or calcium or magnesium ion concentrations (5, 9). Serum binding may greatly affect antibiotic activity (15). In addition, studies by Kunin have demonstrated that rabbit tissue homogenates are capable of binding a number of antibiotics and have special affinity for polymyxin and aminoglycoside antibiotics (6). The present studies demonstrate similar binding with purulent material obtained from humans. Pus was found to bind polymyxin and aminoglycoside antibiotics avidly. The large numbers of binding sites in purulent material was indicated by the high concentrations of antibiotic which could be bound by a single milliliter of pus. The presence of binding rather than destruction of drugs by pus was indicated ANTiMICROB. AG. CHEmarHzR. by the effect of time, temperature, and ph on drug-pus interaction, and by studies showing that the drug-pus interaction's were reversible. Mechanisms of drug binding to tissues are poorly understood. Kunin postulated that electrostatic binding was probably responsible for antibiotic binding to animal tissue homogenates (6). He observed that inhibitory substances in tissue were relatively heat stable and that binding was reversible at least in part. Tissue inhibitors of polymyxin were thought to be acid phospholipids and appeared to bind to particulate cell fractions (6). Aminoglycoside binding appeared to correlate with the number of amine groups present on the molecule and to represent interaction of the drug with the organelles of disrupted cells (6). The abundance of intact cells and cell fragments present in purulent material are probably responsible for binding of aminoglycoside and polymyxin antibiotics to purulent material. The relative insusceptibility to antibiotics of bacteria present in pus was established clearly by the classic studies of Eagle and of Wood and Smith (3, 16). A careful review of those studies reveals that bacteria within abscesses were killed by antibiotics, albeit at a slower rate than bacteria in broth (3, 12). It has not been established whether bacteria at the periphery of abscesses or those present in the wall of abscesses are insusceptible to antimicrobial activity. Therefore, the question of drug binding and drug delivery to such sites may have clinical relevance regardless of the relative inhibitory environment of the abscess milieu. Antibiotic delivery to tissues may be effected by drug diffusion as well as drug binding. Differentiation of the separate effects of diffusion and binding is difficult if not impossible to resolve by in vitro studies and will require in vivo studies. It is of interest that in one instance where gentamicin level of abscess material was tested the drug activity was not detectable (8). However, such studies did not exclude the possibility that drug binding by purulent material had taken place. The composition of purulent exudates is poorly defined. To our knowledge, there is no information concerning the presence or biological significance of endotoxin or penicillinase within abscesses. However, those substances theoretically could be present in abscesses caused by gram-negative bacteria. Since endotoxin has been shown to interact with polymyxin B (11), and since endotoxin and penicillinase production are properties common to a number of gram-negative bacilli, purulent ma-

6 VOL. 6, 1974 terial associated with bacillary infections might be capable of binding or inactivating antibiotics by several means. To exclude that possibility and to simplify interpretation of antibiotic interaction with purulent exudates, the pus from patients with pneumococcal empyema was selected for examination. The present studies suggest that purulent exudates per se are capable of binding polymyxin B, colistin, and gentamicin independent of the presence of factors such as endotoxin and metabolic products derived from gram-negative bacteria. It might be anticipated that purulent material containing endotoxin would bind polymyxin antibiotics even more avidly than shown in the present studies. Pus from patients with Pseudomonas infections was not examined in the present studies because such material is difficult to acquire in sufficient quantity and because the likelihood of polymicrobic infection and variable concentration of endotoxin in such material makes comparison of gentamicin and polymyxin activity difficult. A further difficulty in using purulent material from patients with Pseudomonas infections is the difficulty in sterilizing it (personal observations). Although study of the effectiveness of antibiotics in sterilizing such material is a desirable objective, it was not the goal of the present investigation. Such studies should be done in the future. The clinical relevance of the present studies is uncertain. Perhaps these observations provide an additional reason to the long list of reasons why drainage is the treatment of choice for abscesses. Since carbenicillin is not bound by pus, carbenicillin should be delivered readily to the abscess interior. Therefore, one might hypothesize that carbenicillin or other penicillin congeners might be especially useful in management of patients with abscesses caused by gram-negative bacilli. The validity of this hypothesis must be established by in vivo studies but is consistent with enhanced survival figures observed when carbenicillin is used to treat Pseudomonas infections in patients with leukemia (1). The present studies raise an additional consideration concerning the relevance of gentamicin, colistin, or polymyxin B levels measured in purulent secretions. Although antibiotic assay of the supernatant of purulent material may reflect the effective antimicrobial activity therein, such studies may grossly underestimate INTERACTION OF PUS WITH ANTIBIOTICS 707 the total antibiotic concentration of purulent exudates because of failure to assess binding or other interactions possible between antibiotics and pus. Studies of drug delivery to sites of inflammation should evaluate drug binding to purulent material as a potential source of error. ACKNOWLEDGMENTS This study was supported by Public Health Service grant no. Al from the National Institute of Allergy and Infectious Diseases and by a grant from the Warner-Lambert Research Institute. LITERATURE CITED 1. Bodey, G. P., and V. Rodriguez Advances in the management of Pseudomonas aeruginosa infections in cancer patients. Eur. J. Cancer 9: Bryant, R. E., A. F. Hood, C. E. Hood, and M. G. Koenig Factors affecting mortality of gram-negative rod bacteremia. Arch. Int. Med. 127: Eagle, H Experimental approach to the problem of treatment failure with penicillin. Amer. J. Med. 13: Florey, M. E., E. C. Turton, and E. S. Duthie Penicillin in wound exudates. Lancet ii: Gilbert, D. N., E. Kutscher, P. Ireland, J. A. Barnett, and J. P. Sanford Effect of the concentrations of magnesium and calcium on the in vivo susceptibility of Pseudomonas aeruginosa to gentamicin. J. Infect. Dis. 124:S37-S Kunin, C. M Binding of antibiotics to tissue homogenates. J. Infect. Dis. 121: McCabe, W. R., and G. G. Jackson Gram-negative bacteremia. II. Clinical, laboratory, and therapeutic observations. Arch. Int. Med. 110: McCall, C. E Second intemational symposium on gentamicin, penicillin, and synergism discussion. J. Infect. Dis. 124:S Medeiros, A. A., T. P. O'Brien, W. E. C. Wacker, and N. F. Yulug Effect of salt concentration on the apparent in vitro susceptibility of Pseudomonas and other gram-negative bacilli to gentamicin. J. Infect. Dis. 124:S59-S Moncrief, J. A., and C. Teplitz Changing concepts in burn sepsis. J. Trauma 4: Rifkind, D Studies on the interaction between endotoxin and polymyxin B. J. Infect. Dis. 117: Smith, M. R., and W. B. Wood, Jr An experimental analysis of the curative action of penicillin in acute bacterial infectious. II. The role of phagocytic cells in the process of recovery. J. Exp. Med. 103: Teplitz, C Pathogenesis of Pseudomonas vasculitis and septic lesions. Arch. Pathol. 80: Ungar, J Penicillin in tissue exudates after injection. Lancet i: Verwey, W. F., and H. R. Williams, Jr Binding of various penicillins by plasma and peripheral lymph obtained from dogs, p Antimicrob. Ag. Chemother Wood, W. B., Jr., and M. R. Smith An experimental analysis of the curative action of penicillin in acute bacterial infection. L. The relationship of bacterial growth rates to the antimicrobial effect of penicillin. J. Exp. Med. 103: