A LABORATORY TEST FOR BACTERIAL SENSITIVITY TO COMBINATIONS OF ANTIBIOTICS

A LABORATORY TEST FOR BACTERIAL SENSITIVITY TO COMBINATIONS OF ANTIBIOTICS ERNEST JAWETZ, M.D., PH.D., JANET B. GUNNISON, M.A., VIRGINIA R. COLEMAN, B.A., AND HENRY C. KEMPE, M.D. WITH THE TECHNICAL ASSISTANCE OF CLARENCE R. MIEDEMA, ELIZABETH R. MERRILL, LOIS CHANDLER, AND MARIE STONE Departments of Microbiology, Medicine, and Pediatrics, University of California Medical Center, San Francisco 22, California The increased incidence of resistant strains of microorganisms, following the use and misuse of antimicrobial drugs on a large scale, has led to the administration of combinations of drugs. Physicians often seek advice from laboratory personnel regarding the choice of drugs, and there are increasing numbers of requests to determine sensitivity of organisms to combinations of antibiotics. At present there are no generally accepted methods for such tests, and the purpose of the determination is often poorly defined. We have studied the dynamics of antimicrobial action of combined drugs, 3 ' " 8 and have tried to use such theoretical knowledge in the development, of a relatively simple, inexpensive and comprehensive test suitable for the average clinical laboratory. The proposed method was developed independently from procedures employed in earlier work by our group. 3, 6 ~ 8 We attempted to compromise between ideal and practical aspects by incorporating desirable technical features used by many workers. 1, - 4 5> ' <J " 13 The general theoretical background of sensitivity tests with combined antibiotics was presented in an earlier review. 8 DESIGN OF TEST In order to arrive at a suitable technic the following postulates were made: 1. The microbial population in the test should be large enough to be representative of the population in the host. Therefore, the bacterial inoculum should be 10 6 to 10 8 organisms. This permits the ready emergence of resistant variants, and, consequently, many cultures seem to be more "resistant" than when smaller inoculums are used. 2. The results of the test need not indicate the exact minimal concentration of drug that is bacteriostatic or bactericidal for the microorganism. Inasmuch as concentrations of drugs vary greatly in different tissues or body fluids, as well as at different times, the statement of an exact effectiveness cannot be translated into a meaningful dosage of drug. It is sufficient to know that for a Received for publication March 15, 1955; accepted, June 17, 1955. This study was supported by grants from the Committee on Research, Council on Pharmacy and Chemistry, American Medical Association; Chas. Pfizer and Co.; and the National Institutes of Health (E214). Dr. Jawetz is Professor of Microbiology, Miss Gunnison is Associate Professor of Microbiology, Miss Coleman is Associate in Microbiology, and Dr. Kempe is Assistant Professor of Pediatrics. 1016

Sept. 1955 BACTERIAL SENSITIVITY TO ANTIBIOTICS 1017 given microorganism a drug is: (a) "effective," i.e., the organism is inhibited or killed by low concentrations of drug, readily attainable in the appropriate part of the human body by small or average doses; (b) "slightly effective," i.e., the organism is inhibited or killed only by high concentrations of drug, attainable only by administration of large or even maximal doses; or (c) "ineffective," i.e., the organism is not affected by concentrations of drug attainable by the maximal doses tolerated without serious toxic side-effects in the patient. A drug may be placed in one of these 3 categories after observing the behavior of the microorganism in only 2 concentrations. 12 3. Combined antibiotic effects are probably of little importance clinically, unless they occur in a considerable range of concentrations and ratios of drugs. Earlier work 3 ' 6-8 demonstrated that antibiotic antagonism occurs in a limited zone of concentrations, and such antagonism might not be noted if too few dilutions of drug were employed in the test. In developing a screening test one need not consider antagonism, inasmuch as this phenomenon is probably of little importance clinically. 2 Synergism, on the other hand, occurs in a wide range of concentrations, and it is affected only little by the ratios of drugs in 3, 6-8 the mixtures. Hence, a test including only one or two concentrations of drug is likely to detect this kind of effect by combined drugs. 4. Every available antibiotic need not be tested in all possible combinations. It is justifiable to use only one member of each group of drugs known to give complete cross-resistance and to have a closely similar microbial spectrum and activity. For example, one representative of the tetracycline group is sufficient because differences in behavior in vitro tend to be slight, and other drugs such as polymyxin B may be omitted unless specifically indicated. 5. Combinations for testing may be selected by a tentative empirical scheme based on extensive studies in this laboratory. 7 Antimicrobial agents may be arranged in the following twe'groups: I. Penicillin, streptomycin, bacitracin, neomycin, polymyxin B II. Chloramphenicol, the tetracyclines, erythromycin, (sulfonamides) Synergism is encountered most commonly among drugs of Group I and occasionally when a drug of Group II is added to an agent of Group I, but not among drugs of Group II. If this scheme is accepted as a working hypothesis, then combinations may be limited to I + I and I + II, whereas II + II combinations may be omitted. 6. In order to obtain information on both the inhibitory and the lethal effects of combinations of drugs, a 2-step procedure must be employed: (1) the microorganisms are exposed to the drugs, and (2) they are removed from the drugcontaining environment and tested for viability by subculturing in a drug-free medium. The second step should permit a quantitative estimate of the number of surviving organisms. 7. Simplicity and rapidity of the test, and restrictions in materials and equipment, must take precedence over accuracy, which is limited even when more elaborate tests are done.

1018 JAWETZ ET AL. Vol. 25 OUTLINE OF METHOD Mediums The medium for the first step is Proteose peptone #3 broth (Difco), distributed by an automatic pipetting machine in 10-ml. amounts in standard test tubes (16 by 150 mm.) and autoclaved at 15 lb. for 15 minutes. Aluminum caps, screw caps, or cotton plugs may be used. For certain organisms, e.g., streptococci from patients with bacterial endocarditis, the addition of 5 per cent of blood is desirabie. For the second step, Proteose peptone #3 base (Difco) is used to prepare plates of agar medium with 2 per cent of blood. Phenylethylalcohol (0.25 per cent) also is added to the medium when testing the Proteus species and other rapidly spreading organisms. The mediums have a ph near 7.3. Antibiotics* Stock solutions of antibiotics are prepared once every month and kept at 4. C. During this period there is no noticeable loss of potency of the drugs except chlortetracycline, which must be kept frozen at 20 C. The strength of a stock solution is adjusted so that 0.05'to 0.1 ml. mixed with 10 ml. of broth results in the selected "low" concentration of drug, and 0.2 to 0.5 ml. added to 10 ml. of broth provides the selected "high" concentration. Thus, 2 concentrations of an antibiotic (a 4- to 10-fold range) may be obtained from 1 stock solution with a single 1-ml. pipet, and the accuracy is within 10 per cent. The contents of each tube are thoroughly mixed. Some drugs are available in appropriate amounts in sterile vials ("Antibiotic Testing Kits" of Chas. Pfizer and Co.), and suitable stock solutions are easily prepared by adding a measured quantity of 0.85 per cent solution of sodium chloride. Other drugs are weighed, dissolved in saline solution (erythromycin is first moistened with a drop of ethanol) and filtered through Seitz pads. The strengths of stock solutions currently employed are given in Table 1, together with the approximate final "low" and "high" concentrations obtained by mixing the indicated amount of stock solution of drug with 10 ml. of broth. The number of antimicrobial agents is of course not limited, and additional drugs may be included in the test. The concentrations employed were selected arbitrarily on the basis of estimates of blood level cited by Welch and associates. 14 Owing to the fact that the test is significant only when applied to organisms somewhat resistant to single antibiotics, only relatively high concentrations of drug are included. Sensitivity to combined drugs need not be tested if the bacteria are highly sensitive to individual microbial agents. Performance of the Test The complete test, from receipt of a culture to reporting the final results, requires 72 hours. Intermediate and preliminary results are available at 24 and * We are indebted to Dr. J. W. Smith for supplies of erythromycin hydrochloride, to Dr. E. L. Burbidge for neomycin sulfate, to Dr. N. Bohonos for tetracycline. All other drugs were kindly made available in generous amounts by Dr. F. C. Fink in "Antibiotic Testing Kits" prepared by Chas. Pfizer and Co., Brooklyn, New York.

Sept. 1955 BACTERIAL SENSITIVITY TO ANTIBIOTICS 1019 TABLE 1 PREPARATION OF "LOW" AND "HIGH" CONCENTRATIONS OF ANTIBIOTICS Drug Stock Solution, Mg./ml. Transfer to, 10 ml. Broth (ml. of Stock) Final "Low" Concentration, Mg/ml. Transfer to 10 ml. Broth (ml. of Stock) Final "High" Concentration, Mg/ml- Penicillin Bacitracin Streptomycin Chloramphenicol Tetracycline! Erythromycin Polymyxin B Neomyci n 200* 40* 500 400 400 40 10* 40 0.05 0.05 0.1 0.05 0.05 0.05 0.1 0.05 1 0.2 5 2 2 0.2 0.1 0.2 0.5 0.25 0.4 0.25 0.25 0.5 0.5 0.5 10 1 20 10 10 2 0.5 2 * Penicillin and bacitracin expressed in units/ml. Polymyxin B expressed as equivalent of the weight of pure material. f Oxytetracycline and chlortetracycline, if used, were employed in the same concentrations as tetracycline. 48 hours. The total time required for an inexperienced technician to set up the test, read and record the results, and give the final report is less than 60 minutes for one organism. A sample protocol is given in Table 2. One may assign a number to each drug or combination of drugs, as in the sample protocol, and place each tube in a certain position in a rack. Thus, it is not necessary to label every tube, and only concentrations of drug employed in the combinations need be recorded. Day 1. Broth is inoculated either from an available broth culture or from a culture on solid medium. The inoculum should be sufficiently heavy to give faint opalescence or turbidity. This culture is incubated at 37 C. for 4 to 6 hours, or until the turbidity matches that of the #3 tube of the McFarland nephelometer. From this broth culture, 0.05 ml. is inoculated into 10 ml. of broth containing each of the "low" and "high" concentrations of single antibiotics listed in Table 1, and into 10 ml. of drug-free broth. The tubes are mixed gently, but thoroughly, and incubated overnight at 37 C. Day 2. The tubes set up the previous day are observed for turbidity, which indicates lack of inhibition of growth, and the readings are recorded. Drug combinations are then set up as follows: Group I + I P + B S + N* P + S S + Po* B + S Group P + C P + T P + E B + C B + T B + E + // S + C S + T S + E Po + C* Po + T* N + C* N + T* N + E* * Included only with staphylococci and certain gram-negative rods. Drug symbols are the first letters of the name (see Table 1).

1020 JAWETZ ET AL. Vol. 25 SAMPLE TABLE 2 PROTOCOL ILLUSTRATING THE READINGS AND CONCLUSIONS FROM A TEST FOB SENSITIVITY TO COMBINED ANTIBIOTICS Patient R. B. Date set up: 3-10-53 Referring Physician Dr. B. S. P. Micrococcus pyogenes var. aureus from bacterial endocarditis Set-up on Day 1 No. 1 2 3 4 5 6 7 8 9 10 11 12 13 Drug P 1 P 10 B0.2 B 1.0 S5 S20 C2 C 10 T 2 T 10 E0.2 E2.0 Control Turbidity on Day 2 + + Day 3 I I I + + No. Colonies in Subculture Day 3 200 50 80 60 80 Turbidity on Day 4 + 1 + 1 1 1 No. 25 26 27 28 29 30 31 32 33 34 35 36 37 Set-up on Day 2 Drug P 10 + B 1 P 10 + S 5 B 1 + S5 P 10 + C 2 P 10 + T 2 P 10 + E 0.2 B 1 + C 2 B 1 + T 2 B 1 + E 0.2 S5 + C2 S 5 + T 2 S 5 + E 0.2 Control Turbidity on Day 3 ± + + Number of Colonies in Subculture Day 3 35 8 200 100 100 100 100 50 100 10 200 Confluent bidity on >ay 4 5" + + + + Report Effective single drug: Tetracycline 2 iig./ml. is strongly bacteriostatic, perhaps slowly bactericidal. Effective combinations which are rapidly bactericidal: Penicillin 10 units/ml. + streptomycin 5 /ag./ml. Streptomycin 5Mg-/ml. + tetracycline 2 ^g./ml. Alternates: Penicillin 10 units/ml. + bacitracin 1 unit/ml. Bacitracin 1 unit/ml. + tetracycline 2 /tg./ml. Only one concentration of each drug is employed in the combination. The low concentration is included in combinations if one or both tubes of the single drug remained clear during the first 24 hours. The high concentration is included in the combination if both tubes with a single drug became turbid in the first 24 hours. The combinations are prepared by adding to 10 ml. of broth the same amounts of stock solution used for the test with a single drug, neglecting the small differences in total volume. The combinations and a control tube are then inoculated with 0.05 ml. of culture from the control tube of drug-free broth started on Day 1. All tubes are incubated at 37 C. Day 3. All tubes are examined and presence of turbidity is recorded. One standard loopful (about 0.005 ml.) is removed from all clear tubes (single or combined drugs) and from the drug-free control, and spread onto a sector (one-sixth) of a blood agar plate. These plates and all tubes are placed in the incubator.

Sept. J 955 BACTERIAL SENSITIVITY TO ANTIBIOTICS.1021 Day 4. The presence of turbidity in any of the tubes is recorded. The number of colonies growing on various sectors of the blood agar plates is estimated and ecorded, as in the sample protocol. The test is valid if turbidity is conspicuous in the drug-free broth and numerous colonies develop on its subcultures. If the organisms grow slowly, the final reading of subculture plates should be deferred for an additional 24 hours, and tubes with clear broth may be subcultured a second time on Day 4. Final results that provide an estimate of inhibitory potency and bactericidal effects are usually available within 72 hours. Analysis of Results The selection of "effective single drug(s)" and "effective combination(s) of drugs" is based on a composite evaluation of (a) continued inhibition of growth of the organism as indicated by no turbidity of the broth for 48 to 72 hours, and (b) killing of the organisms as indicated by low colony counts in subcultures on agar. Significant differences in colony counts must be at least 10-fold. Turbidity is recorded on the day the results are observed. In the sample protocol, for example, "turbidity on Day 2" refers to the result after 24 hours of incubation of tubes containing single drugs. Colony counts from subcultures, on the other hand, are recorded for the day that the blood agar was inoculated, e.g., "number of colonies in subculture Day 3" refers to subcultures made on the single drugseries after 48 hours of incubation in broth, but on the combined drug series after only 24 hours of incubation. Thus, for Day 3, similar counts in the 2 series indicate a greater bactericidal effect of the combination. Reports An "effective single drug" is one that inhibits the organism in the low concentration and gives a low colony count on subculture (preferably less than 50). All such drugs are reported so that the physician may select from them. If none of the tested drugs is "effective," those that produce partial or temporary inhibition are reported as "slightly effective." An "effective combination of drugs" is one that results in a tube of broth remaining clear and in a low colony count on subculture (preferably less than 10). If several combinations are approximately equal in effect, they are listed on the report, and the physician may choose from them. The inhibitory and the bactericidal concentrations are stated in order to aid in the selection of drugs and as a guide to dosage. In the sample protocol, tetracycline is an effective single drug as defined above. Among combinations, streptomycin + tetracycline may appear at first glance similar in effect to tetracycline alone, inasmuch as an 8-fold difference in colony count is barely significant. However, the combination permitted only a few organisms to survive after 24 hours' exposure in comparison to 48 hours with tetracycline alone, so that the combination is actually more effective than the single drug.

1022 JAWETZ ET AL. Vol. 25 The physician's decision will depend upon the ease of administering drugs in the necessary doses, the potential toxicities, the distribution of drugs in the body as related to the route of administration, and several other pharmacologic properties and medical considerations. REPRODUCIBILITY OF TEST For purposes of simplicity, the proposed test uses only approximations rather than accurate concentrations of drugs. It was important, therefore, to determine the variability of results obtained by different persons examining the same microbial strain at one time or by the same person performing the test at different times. Thirty-five bacterial strains were submitted to such multiple examinations. On the whole, the agreement was surprisingly good. Disagreements were encountered most frequently in the following situations: 1. Organisms of marginal sensitivity. When the minimal inhibiting dose of a certain drug for a given microorganism happened to coincide with either the low or the high concentration in the test, uncontrollable minor variations might determine whether the organism would be barely inhibited or would grow out rapidly. 2. Emergence of resistant mutants. With some drugs, particularly streptomycin, chance would determine whether a resistant mutant capable of multiplying in a set concentration of drug happened to be present in the inoculum or not. 3. Production of antibiotic-destroying substances. Many strains of staphylococci or gram-negative rods produce penicillinase. The enzyme is present in significant amounts only in cultures incubated for more than 12 hours, whereas cultures incubated less than 8 hours rarely contain enough to inactivate penicillin. Substances that inactivate other drugs may behave similarly. Thus, an inoculum employed on Day 1 (cultured for about 6 hours) might seem sensitive to penicillin, whereas the inoculum on Day 2 (cultured for 18 to 24 hours) might be manifestly resistant to penicillin because of its content of penicillinase. This variable may be controlled by using only inoculums of similar age when testing staphylococci and certain gram-negative rods. In order to estimate the frequency of disagreement in reading duplicate tests on 35 bacterial strains, selected points of comparison were chosen arbitrarily. Comparing the turbidity readings at 48 hours of incubation and the colony counts obtained at 24 or 48 hours of incubation, respectively, significant variations in readings of turbidity occurred in 10 per cent of all tubes, both of single and combined drugs. On the other hand, the colony counts varied by more than 10-fold in an average of 20 per cent of all tubes. The disagreements in duplicate tests performed by the same operator with the same strain of microorganism on different days involved from 5 to 30 per cent of all tubes. The percentage of disagreement in single readings on individual tubes is expected to be higher than in the composite evaluation; the single readings can never give an adequate picture of results in such a complex biologic situation. In the composite readings (turbidity and colony count throughout the period of the test) major disagreements that affected the choice of drugs occurred in only 6 per cent of all tubes.

Sept. 1955 BACTERIAL SENSITIVITY TO ANTIBIOTICS 1023 COMPARISON OF TEST WITH CONVENTIONAL METHODS Can additive or synergistic drug combinations be predicted from the results of sensitivity tests performed with several single antibiotics or will such tests fail to detect useful effects of combined drugs? To answer this question, 138 strains of recently isolated bacteria of varieties frequently drug-resistant were tested simultaneously by standard plate or broth dilution methods with single antibiotics and by the proposed test with combinations of drugs. About 10 per cent of the cultures were entirely resistant to the greatest concentrations of antibiotics employed in the tests, either alone or in combinations. In another 55 per cent of the cultures, the sensitivity tests with single drugs detected one or more agents that were strongly inhibitory or bactericidal. The proposed test demonstrated that one or more combinations of effective single drugs were likewise strongly inhibitory or bactericidal. However, the drug or combination of choice could be predicted accurately from the results with single drugs. With these groups of bacteria the proposed test had no advantage over other conventional methods. In the remaining 35 per cent of cultures, the proposed test detected one or more combinations of antibiotics with strongly inhibitory or bactericidal action that could not be predicted from results with single antibiotics. Tests with single drugs indicated that these bacteria were frequently susceptible to only one antibiotic, which usually caused only temporary inhibition. On the other hand, it was often synergistic with other drugs, resulting in rapid bactericidal action. Such results could be expected on the basis of earlier studies 3,8 on action of combined antibiotics in which it was demonstrated that only one of a pair of drugs exhibiting synergism need show manifest antibacterial action in comparable concentration when acting alone. Some random examples are shown in Table 3. AVith certain bacteria the proposed test for sensitivity to combinations of antibiotics was clearly superior to laboratory examinations that were limited to single antibiotics. In some instances the proposed method had no qualitative advantage over tests with single drugs, but a definite quantitative advantage was noted. For example, the "low" concentrations of 2 drugs in combination maybe completely inhibitory, whereas a 5- to 10-fold larger amount of each drug may be required for inhibition when acting singly. Such information is particularly useful when drugs with a narrow range of safe dosage, such as bacitracin, neomycin, or polymyxin B, must be administered. In such instances it is important to know that one of these drugs may be highly effective in low dosage if it is used with another agent. 3, 8 The conventional serial dilution tests (on solid or in liquid mediums) for sensitivity to antibiotics usually provide information only on inhibition of growth, but not on bactericidal effects. While in many disease processes bacteriostatic activity of the antibiotics is sufficient for cure, this is not always the case. In certain illnesses, such as bacterial endocarditis, osteomyelitis, or bacterial meningitis, optimal therapy demands the use of drugs that are rapidly bactericidal. The ability of the proposed test to determine bactericidal activity is a definite asset in such clinical problems.

1024 JAWETZ ET AL. Vol. 25 TABLE 3 DISCREPANCIES BETWEEN "EFFECTIVE" DRUGS FOUND IN CONVENTIONAL SENSITIVITY TESTS WITH SINGLE ANTIBIOTICS AND IN THE PROPOSED TEST WITH ANTIBIOTIC COMBINATIONS Single Antibiotics* Patie Organism Inhibited but Not Rapidly Killed Not Inhibited byt and Rapidly Killed by Za Lo Da Pi Du Ev Do Ha Di Be Ma Ma Ko De Micrococcus pyogenes M. pyogenes M. pyogenes Streptococcus faecalis Streptococcus salivarius Escherichia coli Aerobacter aerogenes A. aerogenes Salmonella sp. Salmonella typhosa Proteus sp. Proteus sp. Pseudomonas aeruginosa P. aeruginosa C 5, E 0.5 B 2, S 20, C 10 T2, E2 None C 20, E 2 C 10 C 10, T 10 T 10 N 2 C 10, T 10 N 2 C 10, T 10 None None None T 10 P 10, T 10, B 2 P 10 P5, B 1 E2 P 10, S 20, B 2 T 10 P 10, S 20, T 10 B2 Po 1.0, N 2 S20 S 20, C 10, Po 1.0 Po 1.0, C 10 Po 1.0, S 20 P 10, S 20 P 10, C 10, T 10 P 10, C 10 Po 1.0, C 10 Po 1.0, C 10 P 1 + T 10, P 1 + C 2 B 0.2 + C 2 P1+T2, S5 + T2 B 0.2 + E 0.2 P 1 + B 1, P 1 + E 2 P 10 + S 20, P 10 + B 0.2 P 10 + T 2, P 10 + C 2 B 0.2 + E 0.2 P 10 + S 20 P 10 + B 2, P 10 + T 10 Po 0.5 + S 20, Po 0.5 + C 2 Po 0.5 + T 2, S 20 + C 2 S 20 + C 10, Po 0.5 + C 10 S 20 + T 2 Po 0.5 + C 10 Po 0.5 + T 2, N 0.2 + S 20 N 0.2 + T 2 P 10 + S 20, P 10 + C 2 P 10 + T 2 P 10 + C 10, P 10 + T 10 P 10 + C 10 Po 0.5 + C 10 Po 0.5 + C 10, Po 0.5 + T2 * Drug symbols are the first letters of the name (See Table 1). Number indicates/ag./ml. or units/ml. Penicillin and bacitracin are expressed in units; polymyxin B as equivalent of the weight of pure material; other drugs in jugf Only those drugs are listed which entered into combined action. RESULTS OF TEST APPLIED TO 307 FRESHLY ISOLATED BACTERIAL STRAINS Most of these strains were sent to us because preliminary tests (usually with paper disks) indicated relatively high resistance to antibiotics, and initial antibiotic therapy failed to induce a striking response. As shown in Table 4, the largest groups of organisms were staphylococci and coliform bacteria. For 35 bacterial strains (11.4 per cent) the test failed to detect any "effective" single drug or combination. These completely resistant organisms were all gram-negative rods, predominantly Proteus and Pseudomonas species. For 194 strains (63.2 per cent) results of the tests indicated that single drugs were effective. Pairs of drugs and single antibiotics were about equally effective against 101 strains (32.9 per cent). For 78 strains (25.5 per cent) one or more pairs of drugs were unequivocally more effective than any single one. Even the most effective combination of

Sept. 1955 BACTERIAL SENSITIVITY TO ANTIBIOTICS 1025 TABLE 4 THE RELATIVE ANTIMICROBIAL EFFECTIVENESS OF SINGLE DRUGS AND DRUG PAIRS AGAINST DIFFERENT BACTERIA Bacterial Species Total No. Strains Tested Number of Strains of Single Drug More effective than any pair As effective as any pair Drug Pair More Effective Than Any Single Drug No Drug Effective Streptococcus faecalis Streptococcus salivarius Micrococcus pyogenes Proteus sp. Pseudomonas aeruginosa Coliform bacilli* Other gnim-negative rodsf 22 13 96 30 32 90 24 10 9 24 8 7 27 8 5 4 57 3 1 25 6 7 0 15 9 S 32 7 0 0 0 10 16 6 3 Totals 307 93 101 7S 35 * Escherichia coli, Aerobacter aerogenes, Klebsiella pneumoniae, and Paracolobaclrum species. f Salmonella, Shigella, and unidentified species. drugs was not rapidly bactericidal for a few of these strains, but several pairs of drugs were effective (completely inhibitory and rapidly bactericidal) against others, although no single drug had a comparable action. Thus, the proposed test provided useful information regarding combinations of antibiotics to be used against one-fourth to one-third of these relatively resistant organisms. REPORT OF CASES The correlation between the results of the proposed laboratory test and the clinical response to treatment was studied in 23 patients with a variety of infections, and the correlation seemed fairly good. If the patients were treated with a drug or a combination that seemed relatively ineffective according to laboratory tests, their response to treatment was often inadequate. On the other hand, patients had rapid, impressive clinical improvement when they received antibiotics that seemed to be highly effective (singly or in combination) in the proposed laboratory test. The difficulty with such evaluation lies in the general nature of infectious processes seen in these times. Most patients tend to improve either spontaneously or with any form of antimicrobial therapy. To arrive at a conclusive decision about the value of a laboratory test as a guide to therapy, it is necessary to have a patient who (a) has a condition that is unlikely to improve spontaneously, (b) fails to respond to treatment with a drug that appears ineffective by laboratory tests, and then (c) promptly recovers after administration of the drug or combination chosen on the basis of the laboratory test. While this set of circumstances is rare, the following case histories fulfill these criteria and suggest that the proposed laboratory test might aid in the selection of antimicrobial therapy in serious infections caused by antibiotic-resistant bacteria.

1026 JAWETZ ET AL. Vol. 25 Case 1.* R. S., a 43-year-old man, became comatose and developed pneumonia following neurosurgical procedures for the drainage of a subdural hygroma. X-ray examination indicated that the pneumonic process was localized in the middle and lower lobes of the right lung. Coagulase-positive, hemolytic Micrococcus pyogenes var. aureus was isolated from bronchial secretions as the predominant organism. By a disk test, the organism was resistant to all antibiotics except chloramphenicol, neomycin, and erythromycin. The patient failed to respond to treatment with penicillin, streptomycin, a sulfonamide, oxytetracycline, and erythromycin, administered concurrently or in sequence over the period of 2 weeks. He continued to have a fever fluctuating between 38 and 39 C. and was unable to expectorate, so that a tracheotomy had to be performed. The pneumonic consolidation of the right lung remained unchanged. The organism isolated was tested by the proposed method for sensitivity to combinations of antibiotics. "Effective" single drugs were chloramphenicol (10 jig./ml.), oxytetracycline (10Mg/ml.), and erythromycin (2Mg-/ml.), all of which were strongly inhibitory. Bacitracin (1 unit/ml.) was only transiently inhibitory. Among pairs of drugs bacitracin (0.2 unit/ml.) -f erythromycin (0.2 Mg./ml.) was the most effective, being completely inhibitory and rapidly bactericidal. The 2 next best combinations were streptomycin (20 /^g./ml.) + oxvtetracycline (2 /jg./ml.) and bacitracin (0.2 unit/ml.) -f oxvtetracycline (2 Mg./ml.), both of which were only bacteriostatic. On the basis of these findings the patient was treated daily with 40,000 units of bacitracin intramuscularly, 1.5 grams of erythromycin intravenously, and 1 gram of oxytetracycline intramuscularly. The response was rapid and he improved spectacularly. He became afebrile and the lesion in his lung resolved as determined by x-ray film. He was able to expectorate large amounts of purulent sputum. After 17 to 20 days the drugs were discontinued and the patient convalesced uneventfully. Case 2.f R. S., a 42-year-old woman with rheumatic heart disease since adolescence, developed bacterial endocarditis. A coagulase-positive, hemolytic Micrococcus pyogenes var. aureus was cultured repeatedly from her blood. This organism seemed to be moderately resistant to several antibiotics. The patient received several courses of antibiotics during a period of 7 months, including penicillin (up to 30 million units daily by intramuscular drip) and combinations of oxy tetracycline with penicillin, oxytetracycline with carbomycin, and erythromycin with penicillin at various times. While receiving some of these antibiotics the patient improved, but she always relapsed clinically, and the organisms were isolated again, as soon as the drugs were discontinued. The staphylococcus was inhibited, but not rapidly killed in vitro by penicillin (0.1 unit/ ml.), bacitracin (1 unit/ml.), streptomycin (12.5 ^g./ml.), chloramphenicol (4 jug./ml.), oxy tetracycline (4 jug./ml.), or erythromycin (0.4 Mg./ml.). Among 12 drug combinations examined, onh' the following were rapidly bactericidal: streptomycin (5Mg-/ml.) + oxytetracycline (2 Mg./ml.), and bacitracin (0.2 unit/ml.) + oxytetracycline (2 jug./ml.). On the basis of these results, the patient was placed on a daily regimen of 2 grams of oxytetracycline orally and 1.5 grams of streptomycin intramuscularly in divided doses. She received these drugs for 4 weeks, and clinical response was rapid. The patient became afebrile, previously marked anemia subsided, and she gained weight and strength. Blood cultures during the next 3 months yielded no growth of staphylococci. Case S. J. W., a 42-year-old Negro man, entered the hospital with an acute febrile illness of 5 days' duration, characterized by agonizing pain of the right hip, headache, and a productive cough with a cupful of sputum daily. The patient was a narcotic addict and had many scars suggesting self-medication with narcotics. A sj'stolie heart murmur of changing quality was heard over the precordia. An x-ray film of the chest revealed patchy densities in the right lung field. Repeated blood cultures were positive for coagulase-positive, hemolytic Micrococcus pyogenes var. aureus. The tentative diagnosis was septicemia caused * Reported by permission of Dr. P. Pillsbury. t Reported by permission of Dr. S. D. Leo.

Sept. 1955 BACTERIAL SENSITIVITY TO ANTIBIOTICS 1027 by Micrococcus pyogenes, with pneumonitis, osteomyelitis of right hip, and probably endocarditis. Initial therapy with penicillin and tetracycline in full systemic doses had no significant effect. Readings from the proposed test indicated that the organism was inhibited, but not rapidly killed, by streptomycin (20 /ig./ml.), terramycin (10 ^g./ml.), or erythromycin (2 Mg./ml.). It was not affected by penicillin (10 units/ml.) or bacitracin (1 unit/ml.), and was only briefly inhibited by chloramphenicol (10 jug./ml.). Among drug combinations the following pairs were completely inhibitory and rapidly bactericidal: streptomycin (2 Mg./ml.) + erythromycin (0.2 /ig./ml.), and streptomycin (20 /»g./ml.) + oxytetracycline (2 M g./ml.). On the basis of these findings, streptomycin + erythromycin might have been chosen because lower concentrations of drug were required than with the other combination. However, the patient was treated daily for 2 weeks with 2 grams of streptomycin intramuscularly, 1.2 grams of oxytetracycline intramuscularly and 1 gram orally. The temperature gradually returned to normal, with occasional spikes of fever. While blood cultures remained negative on this regimen he was subsequently placed on 2 grams of streptomycin intramuscularly and 2.4 grams of erythromycin orally, daily for 4 weeks, and more rapid improvement occurred. The lesion in the lung cleared, the hip improved greatly, and the blood cultures remained negative for 6 weeks after therapy. Case 4* F- A., a 6-year-old girl, had recurrent bouts of infection of the urinar\ r tract for more than 3 years. Symptoms and clinical and laboratory examinations suggested chronic pyelonephritis with recurrent exacerbations and involvement of the lower urinary tract. Repeated urologic study failed to demonstrate any obstruction. Bacteriologic examinations revealed from 10,000 to 10 million Escherichia coli per ml. of urine at various times. The patient received more than a dozen courses of chemotherapy with a variety of drugs, but had no lasting improvement. A specimen of catheterized urine obtained on March 31, 1954, during a quiescent period, revealed 14,000 E. coli per ml. of urine. According to results of the proposed method, "high" amounts of penicillin, streptomycin, bacitracin, erythromycin, neomycin or polymyxin B failed to inhibit the organism. Chloramphenicol or tetracycline (10 jug./ml.) inhibited the organism temporarily but failed to kill it. "Effective" combinations of drugs that were inhibitory and rapidly lethal to the organism were: polymyxin B (0.5 Mg-/ml.) + tetracycline or chloramphenicol (2pg./ml.), and streptomycin (5 /ug./ml.) + chloramphenicol or tetracycline (2yug./ml.). Accordingly, the patient was treated from April 9 to April 20, with daily doses of 50 mg. of polymyxin B intramuscularly, 250 mg. of streptomycin intramuscularly, and 300 nig. of tetracycline orally, in divided doses. Cultures at the end of treatment yielded growth of only a few streptococci of the viridans type. Fourteen to 15 months later cultures of the urine yielded no growth, and the patient was apparently well. Case 5. A. M., a 71-year-old man, had frequency and nocturia for a year, and a transurethral resection was performed 2 weeks prior to entry in order to relieve obstruction resulting from benign hypertrophy of the prostate gland. At that time his urine contained many bacteria and leukocytes, and he received penicillin and streptomycin postoperatively for a week, until discharged from the hospital. His postoperative course was not remarkable until 3 days prior to entry, when he had a sudden onset of chills and fever, and soon developed circulatory collapse. When admitted, he was in shock, and the blood pressure was 50/10. No localizing signs were found, and blood cultures were made because of a suspected bacteremia. The patient was given penicillin, streptomycin, and tetracycline intravenously, in addition to dextrose and norepinephrine intravenously. He improved somewhat for 24 hours, but then relapsed into shock. Proteus mirabilis was isolated from 2 blood cultures, and results of the disk test indicated that the bacteria were completely resistant to antibiotics and sulfonamides. The organisms * Reported by permission of Dr. D. R. Smith.

1028 JAWETZ ET AL. Vol. 25 were tested by the proposed technic, but, in view of the urgency of the situation, not only the single drugs but also the combinations of drugs in "high" concentration were set up immediately. While none of the single drugs in "high" concentration was significantly inhibitory, the combination of penicillin (10 units/ml.) + chloramphenicol (10 jug./ml.) was completely inhibitory and fairly rapidly bactericidal. Accordingly, the patient was given 15 million units of penicillin and 1.5 grams of chloramphenicol intramuscularly each day, starting 48 hours after entry. He also received a transfusion of whole blood. Within 24 hours there was definite improvement in clinical appearance and the blood cultures yielded no growth. Subsequently, the patient had a relatively rapid and uneventful recovery. The antibiotics were discontinued 10 days later and the patient was discharged, evidently cured of his bacteremia due to P. mirabilis. Cultures of the urine a week later had growth of a few Streptococcus faecalis and Escherichia coli, but no Proteus organisms were found. DISCUSSION In the majority of bacterial infections the administration of a single antimicrobial drug in adequate doses for a sufficient length of time results in cure of the patient and eradication of the infecting microorganism. When a specific microorganism is identified as the cause of the illness, the most suitable drug can frequently be chosen on the basis of past experience and antibiotic sensitivity tests are unnecessary. 5 This is true particularly for infections caused by microorganisms that are highly sensitive to drugs, e.g., group A beta hemolytic streptococci, pneumococci, meningococci, and Hemophilus influenzae. When such infections are encountered, the physician is rarely justified in requesting any antibiotic sensitivity test. A minority of clinical infections is caused by microorganisms that are either poorly inhibited or not readily killed by the common antimicrobial agents in average doses. It is in these relatively few selected cases that antibiotic sensitivity tests are indicated, and the results may be a guide to successful therapy. A request for "culture and sensitivity tests" on every specimen will often cloud the issue of selecting a drug, rather than clarify it. Routine laboratory tests cannot be substituted for clinical judgment. The "informed guess" of the experienced clinician is often more valid than results from nonselective laboratory examinations. The "validity" of antibiotic sensitivity tests has been contested by proponents of the various methods, each claiming for his test reliability, simplicity, reproducibility of results, and close correlation with clinical experience. Absolute correlation of the in vitro result with clinical response, however, should not be expected. In the laboratory test there is a simple interaction between drug and microorganism, influenced only by the physical and chemical environment which can be controlled to a large extent. In the patient, on the other hand, the body enters into highly complex interactions with the drug and the microorganism, often resulting in drug-parasite relationships entirely different from those in vitro. In combined action of antibiotics, the situation is even more complex owing to the many possible interactions of drugs and the ways in which a mixture of drugs might influence the infection. The evidence presented suggests that the proposed laboratory test is as com-

Sept. 1955 BACTERIAL SENSITIVITY TO ANTIBIOTICS 1029 prehensive and useful in selected cases as any other sensitivity test now available. It has the advantage of flexibility, permitting evaluation of the inhibitory and the lethal properties of single or multiple drugs over a period of time that includes the early and later effects of drugs. Inasmuch as the observation period can be prolonged, it is possible to perform useful tests with slow-growing organisms. On the other hand, if the urgency of the clinical problem demands, the test may be accelerated (as illustrated in Case 5). Thus, the combinations of drugs may be set up in "high" concentrations on Day 1 when dealing with presumably highly resistant organisms. If an apparently "effective" combination is found, treatment with large doses of these drugs may be started without delay, and, simultaneously, additional combinations may be tested, based on the results with single drugs. This may permit a subsequent readjustment of the doses of drugs if the initial response of the patient was encouraging. Many such minor modifications in technic are possible if the basic principle of the test is kept in mind. Likewise the bacteriologic mediums may be varied. In the case of fastidious organisms, e.g., certain streptococci from bacterial endocarditis, it is desirable to add sheep or rabbit blood (in a concentration of 5 per cent) to the 10-ml. broth tubes. Anaerobic bacteria can be tested by incubation in anaerobic jars. In spite of the ability of thioglycollate medium to inactivate several antibiotics, satisfactory tests have been performed with some anaerobic bacteria in thioglycollate medium in 10-ml. amounts. The relatively high concentrations of antibiotics apparently were sufficiently stable in thioglycollate. Mixed infections of wounds, body cavities, and abscesses are notorious problems in treatment. It is necessary to consider not only the effect of antimicrobial drugs on each of the participating microorganisms, but also the effect of the organisms on each other. Because the rate of growth of organisms in vitro is quite different from their behavior in vivo, the testing of antibiotic sensitivity with mixtures of bacteria tends to be unreliable. In spite of these drawbacks a number of tests were performed by the proposed method on mixtures and the individual strains of organisms isolated from mixed infections. Unexpectedly, the results were found to be comparable. Thus the interaction of different bacterial species in vitro evidently did not influence the over-all results enough to give a misleading impression of antibiotic activity in these particular instances. The proposed test has been used in this laboratory for more than 30 months. We are satisfied that it is as reliable and informative as other procedures used for this purpose, and that it requires less material, time, and effort. Furthermore, it is our opinion that this test actually provides more useful information and clinical guidance than other procedures requiring the same amount of time and materials. Its chief drawbacks are the lack of a rigid, inflexible protocol for its performance and the necessity of an understanding, intelligent interpretation of the composite readings. Hence, the test is not suitable for a "routine" applied to any bacterium isolated in the clinical laboratory. It is used to best advantage with microorganisms known to tend toward antibiotic-resistance, isolated from patients who either have a serious disease or have failed to respond to initial chemotherapy.

1030 JAWETZ ET AL. Vol. 25 The test must be used with judgment and its interpretation requires even greater judgment. Only under these circumstances can it be expected to give useful results. SUMMARY A comprehensive test for antibiotic sensitivity must measure both bacteriostatic and bactericidal ability of single drugs or combinations acting on a large bacterial inoculum. Accurate determination of minimal antibacterial concentration of drug is not essential. A 2-step test is proposed which is comprehensive and adaptable: (1) Microorganisms are exposed to antibiotics in liquid medium for varying periods and growth or inhibition is noted. (2) The number of surviving microorganisms is estimated by semi-quantitative subculture onto solid medium. This test requires a minimum of materials and less than an hour of the technician's time. Final results are available 72 hours after the organism is obtained from the patient, but earlier preliminary readings are useful in guiding therapy. The reliability of the test and its reproducibility are believed to be adequate. The proposed laboratory test for bacterial sensitivity to combinations of antibiotics was compared with conventional dilution tests. Among 138 strains of frequently drug-resistant bacteria, 10 per cent were entirely resistant, 55 per cent were sensitive to one or more drugs that could be detected equally well by conventional tests or by the proposed test, and 35 per cent were sensitive to antibiotic combinations that could not have been detected by the conventional tests with single antibiotics. When applied to 307 freshly isolated strains of bacteria, the proposed test was unequivocally superior in detecting effective drug combinations in about 25 per cent of the cases. Five case reports are presented to illustrate the usefulness of the test in selected clinical situations. It is stressed that its flexibilty and comprehensiveness may give the test marked advantages over other, conventional procedures, but these features may also be drawbacks, because they require intelligent interpretation and judgment and prevent the blind, mechanical application of the test. However, when applied with judgment and used with understanding, the test may provide valuable guidance for physicians in the treatment of infections that are relatively resistant to therapy with antibiotics. REFERENCES 1. BRYSON, V., AND SZYBALSKI, W.: Microbial selection. Science, 116: 45-51, 1952. 2. CHABBERT, Y.: Technique simplifiee pour l'6tude de l'action bact6ricide des associations d'antibiotiques. Ann. Inst. Pasteur, 85: 122-125, 1953. 3. GUNNISON, J. B., SHEVKY, M. C, BRUFF, J. A., COLEMAN, V. R., AND JAWETZ, E.: Studies on antibiotic synergism and antagonism: The effect in vitro of combinations of antibiotics on bacteria of varving resistance to single antibiotics. J. Bact., 66: 150-158, 1953. 4. HILSON, G. R. F., AND ELEK, S. D.: Routine testing of the bactericidal action of antibiotics for clinical purposes. J. Lab. & Clin. Med., 44: 589-594, 1954. 5. JACKSON, G. G., AND FINLAND, M.: Comparison of methods for determining sensitivity of bacteria to antibiotics in vitro. Arch. Int. Med., 88: 446-460, 1951. 6. JAWETZ, E.: Antibiotic synergism and antagonism: A review of experimental evidence (Almroth Wright Lecture). Arch. Int. Med., 90: 301-309, 1952. 7. JAWETZ, E., AND GUNNISON, J. B.: Studies on antibiotic synergism and antagonism: A scheme of combined antibiotic action. Antibiotics & Chemother., 2: 243-248,1952.

Sept. 1955 BACTERIAL SENSITIVITY TO ANTIBIOTICS 1031 8. JAWETZ, E., AND GUNNISON, J. B.: Antibiotic synergism and antagonism: An assessment of the problem. Pharmacol. Rev., 5: 175-192, 1953. 9. LEDERBERG, J., AND LEDERBERG, E. M.: Replica plating and indirect selection of bacterial mutants. J. Bact., 63: 399-406, 1952. 10. MARTIN, R., SUREAU, B., AND CHABBERT, Y.: Int6ret au coursdesmaladiesmicrobiennes de l'eiude du pourvoir bact6ricide des antibiotiques sur les germes. L'association antibiotique de choix. Bull. Soc. med. h6p. Paris, 68: 1192-1201, 1952. 11. MAY, J. R., AND MORLEY, C. W.: The routine testing of the sensitivity of bacteria to antibiotics. Lancet, 1: 636-638, 1952. 12. RANTZ, L. A., AND RANTZ, H. H.: Sensitivity of various clinically important bacteria to four antibiotics. Stanford M. Bull., 2: 183-187, 1953. 13. STREITFELD, M. M., AND SASLAW, M. S.: A strip-gradient method for in vitro assay of bacterial sensitivity to antibiotics paired in various concentration ratios. J. Lab. & Clin. Med., 43: 946-956, 1954. 14. WELCH, H., AND OTHERS: Principles and Practice of Antibiotic Therapy. (Medical Encyclopedia Series) New York: The Blakiston Co., 1954, 699 pp.