Antibiotics in Laboratory Medicine, 6 Ed.

Chapter 3. Susceptibility Testing of Antimicrobials in Liquid Media

Daniel Amsterdam

Fleming’s serendipitous observation that the action of Penicillium notatum could repel several species of bacteria became what was to be known as the chemotherapy era. From that discovery, Fleming developed two approaches for assessing antimicrobial activity that are currently used. The initial discovery was determined by an agar diffusion methodology. Subsequent experiments that attempted to characterize this newly found antibacterial substance can be recognized as exemplifying the broth dilution method for susceptibility testing (1). One method Fleming used to obtain quantitative assessments of the degree of activity of an antimicrobial agent was to inoculate a suspension of the organism into a liquid growth medium that incorporated serial twofold dilutions of the agent. This we recognize as the broth macrodilution method. Obviously, it is not the broth that is diluted. The potentially active compound is the component that is diluted. As penicillin became available for therapeutic use, other medical microbiologists quickly adopted this procedure in order to guide therapy (2). No doubt another impetus for antimicrobial susceptibility testing became manifest as a result of financial considerations. Before World War II, the production of penicillin was limited and extremely expensive. Thus, it was evident that a procedure capable of predicting when the use of penicillin would be effective in a particular infectious disease needed to be developed. In the 1940s, several antibiotics were discovered. Although the broth susceptibility method was the first procedure developed for assessing the in vitro efficacy of antimicrobial agents (and it still serves today as a reference method), it was replaced by methods using antibiotic-impregnated filter paper strips (3) and, later, disks (4).

Toward the end of the 1950s, the status of antimicrobial susceptibility testing was in disarray as a result of the lack of acceptable standard procedures. To remedy this situation, an international group of experts was convened. They outlined general guidelines and reported on the need for standards (5). After the World Health Organization (WHO) report, an international collaborative group was formed to address the problems of standardization. The report by this group (6) was later followed by adoption of the procedures of Bauer et al. (7) in the Federal Register (8,9), which in large part was directed toward agar disk diffusion susceptibility testing. The International Collaborative Study (ICS) report (6) was the springboard for establishing a database and objectives of standardization for current methods (9–11).

As it evolved from Fleming’s pioneering studies, the technique of using serial twofold dilutions in a liquid medium for studying the antimicrobial action of therapeutic agents is referred to as the broth dilution method, although the component diluted is clearly the antimicrobial agent. Initially, this technique was performed in test tubes in a final volume of 1 to 2 mL, and most laboratories followed the procedure detailed by the ICS (6). As the number of antimicrobial agents for testing increased, a more manipulative assay was automated and became popular. This procedure, referred to as the microdilution techniquebecause the final volume was only 50 µL, has gained much wider acceptance than the earlier dilution methodology, known as the macrobroth or macrodilution technique.

Commercial development and application of the miniaturized technique has made it accessible to all laboratories. Thus, minimal inhibitory concentrations (MICs) can be provided by any laboratory regardless of its technical resources. Microdilution test results have been nearly equivalent to those obtained with the conventional macrodilution method, with little exception. For gram-negative microorganisms, the microdilution procedure yields MICs approximately one dilution lower than does the macrodilution method (see “Standard Broth Dilution Procedures”).

The focus of this chapter is on selected elements in the design and performance of standard methods and new rapid, automated, and instrumental approaches to antimicrobial susceptibility testing in liquid media. Although reference is made to a variety of fastidious microorganisms, anaerobes and mycobacteria are not discussed (see Chapters 4 and 5).

Test Procedure

In selecting a procedure to determine the outcome of the interaction of microbe and antimicrobial compound, the laboratory must first determine whether qualitative or quantitative information is desired or required. Although, as stated, the central focus of this discussion is on the interaction of bug and drug in a liquid (broth) environment, a comparison of broth and agar diffusion methodologies is warranted (Table 3.1).

It is evident that a distinct advantage of broth methods is that they permit the determination of a minimal bactericidal concentration (MBC) end point. For those laboratories involved in special clinical pharmacologic studies, the assessment of the clinical efficacy of antimicrobial compounds (older or newly formulated); detailed assays of drug levels, distribution, and toxicity; and fully quantifiable results determined by MIC end points derived from broth dilution studies are required. The MIC can in turn be used to calculate therapeutic ratios, that is, the ratio of serum (or tissue) fluid concentration to MIC. The MIC can be determined from standard quantitative broth (or agar) dilution methods and derived from regression analysis of the diameter of zone inhibition (12,13) (Table 3.1). If the laboratory has established a database and has experience with one methodology, changing procedures for an alternative method can create new problems.

The advantages of one approach over another are summarized in Table 3.1. In certain clinical therapeutic situations (e.g., in cases of endocarditis or osteomyelitis), MICs may be indicated. When MICs are obtained, they can be used to calculate therapeutic regimens to minimize the adverse effects of potentially toxic drugs, giving added confidence in the selected therapeutic regimen. If broth is used, the simultaneous or sequential determination of MIC and MBC values can be achieved or the killing rate determined (14). Furthermore, the quantitative assay of drug combinations to detect synergy is more readily achievable in liquid systems and may be required in certain clinical situations (e.g., in cases of infection with Enterococcus).

For clinical laboratories, combining two approaches could prove to be more useful and cost-effective. For example, antimicrobial susceptibility testing of urine samples could be done by agar disk diffusion, and broth dilution could be reserved for isolates recovered from putatively sterile compartments (blood, spinal fluid, synovial fluid, etc.). With this strategy, isolates recovered from nonsterile compartments would not require quantitative studies.

Results obtained from broth dilution susceptibility testing may be less than optimal with certain antimicrobial–microorganism combinations. The bacteria that have produced difficulties in the past decade are Enterococcus, methicillin-resistant Staphylococcus aureus (MRSA), Pseudomonas aeruginosa, Haemophilus influenzae, and Neisseria gonorrhoeae. Antimicrobials that delivered unreliable data included the sulfonamides, trimethoprim, nitrofurantoin, and newer combination drugs as the β-lactam–β-lactamase inhibitors. These problems were attributable to lack of appropriate standardization of inoculum and medium and have been resolved.

Several variables must be controlled to obtain accurate reproducible results. As with any assay system, deviation from standard procedure modifies the results obtained. Table 3.2 outlines some elements of the broth dilution test that have produced variable results. These are discussed in the pages that follow.

Limitations

One must recognize that the laboratory brings together in an artificial way (in vitro) the bug and drug in a setting outside of the host environment. Whether liquid (broth) or semisolid (agar) environments more truly represent the living human milieu is a subject of ongoing discussion and controversy (15). More pointedly, is the microorganism selected for study representative of the infectious agent offending the host? Because nosocomial infections predominate in hospital settings, a key question is whether the microorganisms recovered from patient specimens are representative of infection or colonization. Usually, this question cannot be answered, but it is clear that the most likely pathogen from a specific specimen should be tested and that routine testing of mixed flora should not be done. The identity of the isolate to be tested frequently dictates or influences the methodology. For certain microorganisms, such as Streptococcus pyogenes and Neisseria meningitidis, testing may not be necessary because the outcome can be readily assured using certain antimicrobial agents or the method and interpretative standard are not yet fully established for the species. For certain rare pathogens or microorganisms with slow growth potential, clearer end points may be obtainable in broth as compared with agar.

In the final analysis, it is necessary to realize that in the laboratory setting, there is a continuous exposure of 10⁵ to 10⁶ colony-forming units (CFU) of microorganism per milliliter to a static (albeit minimally varying) drug concentration during the entire incubation period. These conditions do not prevail in vivo, where larger (or smaller) numbers of bacteria at the infected site are exposed to fluctuating drug concentration gradients.

Selection of Antimicrobial Compounds for Evaluation

As the number of approved compounds continues to increase (although now at a limited rate), the challenge for the laboratory is to select rationally a limited number of antimicrobial agents for testing. To test all licensed drugs would be impractical and, in this era of constrained laboratory budgets, uneconomical. In the past, the selection of agents for testing was aided in part by the U.S. Food and Drug Administration (FDA) as it recognized class compounds or, as they apply to agar disk diffusion testing, class disks. However, as the number of chemical classes and congeners within each class increases, the unique pharmacokinetic (PK) properties frequently dictate separate disks for each newly approved compound. With the current proliferation of newer β-lactam agents and quinolones, the class idea of drug selection has become even more important for achieving economy of testing, especially given the availability of the several commercial systems.

Each laboratory and clinical setting should develop a strategy for testing compounds based on previously accrued information on resistance patterns of class agents. Patient demographics as well as patterns of antibiotic use (and abuse) will assist in this selection process. Guidelines to assist laboratories in this process have been proposed by the Clinical and Laboratory Standards Institute (CLSI), formerly named the National Committee for Clinical Laboratory Standards (NCCLS) (10,11) and are included here for convenience (Table 3.3).

It should be noted that the represented groups of organisms are those for which it is most difficult to predict, with any degree of certainty, a susceptibility result based on prior experimental data. The agents in the primary and secondary groups achieve peak levels in serum, whereas those noted in the urine group attain maximum concentrations in that compartment. Final drug selection should represent a consensus opinion at each site or medical center and should include input from the clinical microbiologist, infectious disease specialist, and clinical pharmacist. As noted, these decisions require information about the antimicrobial nature (toxicity and pharmacokinetics) and the probability of resistance for each drug–bug pair previously tested. Acquisition and use costs also need to be considered. Frequently, these data are incorporated into a quarterly or annual summary report distributed by the institution.

Although costs and test constraints may limit the number of agents tested (usually 8 to 12), hospital and/or laboratory information systems can specify the distribution of reports of susceptibility test results to the patient’s chart. In a sense, the problem of selecting agents for testing prompts the following question: What information about which agent should be routinely reported? Compounds that are on restricted or limited formulary status need not be routinely reported, but the data can be stored for consultation and future consideration of routine testing and reporting. The laboratory may also wish to consider testing those agents that are soon to be approved by the FDA, in anticipation of acquiring a database for setting new priorities.

In the performance of any susceptibility assay, the laboratory needs to define its objectives clearly. These should include the following:

 1.  To test organisms recovered from significant sites (e.g., blood or cerebrospinal fluid) to provide a guide for rational therapy.

 2.  To evaluate the susceptibility of selected nosocomial agents to determine variations in resistance patterns.

 3.  To determine and monitor antibiotic susceptibility (resistance) patterns as epidemiologic markers and thus as evidence of the effective use of antibiotics for coverage.

 4.  To study the activity of recently approved and introduced agents as well as experimental agents.

For routine clinical laboratories, an integrated approach is prudent. This approach serves to discourage the potential abuse of newer agents and thus can help minimize the selective pressure on nosocomial isolates, limiting their potential for developing resistance (Table 3.3).

Several caveats need to be mentioned in regard to Table 3.3, which represents a guide for general test selection of antimicrobial agents against selected bacterial genera.

■  For bacteria recovered from cerebrospinal fluid, cefotaxime and ceftriaxone should be tested and reported. Several antimicrobial agents and types may not be effective for treatment. These include the following:

■  Agents administered orally

■  First- and second-generation cephalosporins (except cefuroxime)

■  Clindamycin

■  Macrolides

■  Tetracyclines

■  Fluoroquinolones

■  Rifampin should not be used as the sole antimicrobial agent.

■  The macrolides and clindamycin should not be routinely reported for microorganisms recovered from the urinary tract.

Concentration Range for Testing

When susceptibility plates are prepared in-house or ordered from a commercial supplier, specific concentrations must be selected that conform to the microtiter configuration (conventionally, 8 × 12 wells) or the containers being used. The concentrations selected must satisfy several criteria.

 1.  The concentrations should extend over the end points of a large series of isolates, taking into consideration whether the distribution of end points is dichotomous, unimodal, or bimodal. For example, Streptococcus pneumoniaeand other streptococci (e.g., Streptococcus viridans) are exquisitely susceptible to penicillin. For these organisms, limiting the lower end of the concentration range to 0.5 or 0.1 µg/mL may fail to detect minor but significant variations in the susceptibility pattern. Although such shifts may be minor, they provide a way of tracking emerging resistance. The development of emerging resistance of S. pneumoniae to penicillin has been documented (16,17) but has limited clinical relevance, since the large doses of penicillin that can be administered to patients will kill pneumococci with an MIC 10 times higher than the normal.

 2.  The concentrations tested should include and exceed (by at least one dilution step) the highest concentration found in biologic fluids. For those compounds that are concentrated in the urine in their active form, such levels can be quite high. PK data for compartmentalized antimicrobics in specific compartments must be reviewed when making these selections. Similarly, for potentially toxic agents, such as the aminoglycosides, it is pointless to test expanded upper ranges of concentrations beyond one or two dilutions above the pharmacologically toxic level. Here is one example of the failure of the twofold serial dilution schedule. Because of the narrow safety margin (therapeutic index) of aminoglycosides, rather than increasing the range above (in this case) the urinary level and taking the drug outside the therapeutic index range, it would be more prudent to prepare small arithmetic incremental dilutions for gentamicin (as well as tobramycin) in the 6 to 12 µg/mL range (18) as follows: 0.5, 1.0, 2.0, 4.0, 6.0, 8.0, 10.0, 12.0, and perhaps 16.0 µg/mL. For amikacin, a similar stepping range from 1.0 to 32 µg/mL is appropriate. Table 3.4 fails to note these figures, because the panels are not commercially available and are difficult to prepare. However, they are presented here for the reader’s consideration.

 3.  The range of concentrations included in the panel should permit the end point detection of quality control (QC) strains (see “Quality Control”). Although routine plates and tubes cannot be prepared in anticipation of predicting synergy-checkerboard patterns, it is worth noting that certain species, such as those in the genus Enterococcus, can be readily treated with a combination of penicillin G or ampicillin and an aminoglycoside, usually gentamicin, to produce synergistic bactericidal activity. Testing for synergistic and bactericidal activity is discussed elsewhere in this volume. Although dilution and synergy or killing curves in broth underscore a conventional basis for quantitative methods of testing synergy, such tests represent a special objective outside the scope of standard broth dilution procedures for routine susceptibility tests. However, the enterococci can be challenged with a single aminoglycoside concentration to determine whether the organisms demonstrate high-level resistance and would be susceptible to the synergistic action of a β-lactam and an aminoglycoside (19). Such high-level resistance can be determined with a single concentration outside of the clinically useful range included in Table 3.4. A 500- to 2,000-µg/mL concentration of aminoglycoside (usually gentamicin) could be included to screen for this type of resistance to synergy in the enterococci, as is common in some commercial systems. It is wise to consider whether this single high-level dilution should be added to the standard series or whether it should be part of a separate screening process for all gram-negative bacilli.

It is difficult to arrange a series of concentrations that satisfy all of the criteria noted earlier and that are applicable in the clinical laboratory setting. However, Table 3.4 presents a suggested outline of concentration ranges designed to approach the goals. Two series ranges are indicated in the table. The first (denoted by “X”) represents a basic set of eight dilutions that cover minimum concentrations. These usually include MIC breakpoints suggested for assigning a general interpretive result to three or four categories, regardless of the terminology used to define the interpretive scheme. The use of widely spaced, selected, screening dilution steps frequently employed for certain drugs in commercial microdilution trays should be avoided. The use of widely spaced (skip-step) dilutions results in a less accurate quantitative assay than does a continuum of concentrations (see “Interpretive Guidelines for Susceptibility or Resistance”) and fails to detect upward trending of resistance. Also noted in Table 3.4 are extended concentrations (denoted by “+”) for expanding the basic dilution series to provide additional relevant concentrations. These additional concentrations are added to either the upper or lower end of the range to permit better estimation of possible clinical utility against partially resistant strains in uncomplicated urinary tract infections, where high levels of drugs may be achieved in the urine (20). The addition of the end dilutions (concentrations) allows detection of modest changes in the susceptibility patterns frequently seen with highly susceptible isolates (as mentioned in the preceding section) for organisms such as S. pneumoniae or N. gonorrhoeae or occasional gram-negative bacilli that might be especially susceptible to newer agents. If the laboratory determines that the scheme outlined in Table 3.4 is unacceptable and represents a far too customized situation, a compromise standard series of dilutions can be prepared, as outlined by the ICS (6). Table 3.5 lists dilution concentrations recommended for preparation of standard twofold dilutions in 13 × 100-mm tubes. The scheme represented in Table 3.5 can be extended to include as many end points as desired. As indicated, 14 dilution steps are included, making the scheme broad enough to address most of the problems noted earlier, but it may not be practical or necessary for the average clinical laboratory. For larger volumes of work, when more tubes are necessary, the scheme outlined in Table 3.6 is recommended (6). Here again, although the series may be customized for specific bacterial agents, employing the same standard dilution for all drugs has the advantage of simplicity and utility for routine use.

Preparation and Storage of Stock Solutions

The decision to purchase commercial test systems or to prepare in-house panels depends on the laboratory type (clinical, developmental, etc.) and the objectives. Prior to the development of commercial antimicrobial susceptibility test systems, there was little choice, except for deciding whether agar or broth dilution would be used. In either case, rigorous standards have to be followed for controlling the preparation and maintenance of stock solutions of antimicrobial agents.

Reagent-quality antimicrobial powders can be obtained from the pharmaceutical manufacturer, the FDA, or the U.S. Pharmacopeia (Rockville, MD). Drug preparations stocked by pharmacies for clinical administration should not be used in the laboratory because they may contain preservatives and may not be standardized as carefully as assay-quality powders. Additionally, some clinical preparations (e.g., chloramphenicol sodium succinate) require hydrolysis in vivo to be in the active form.

Laboratories need to maintain a detailed registry of the antibiotic powders requested that indicates the date of receipt, supplier, lot number, assay potency, outdate, and storage conditions and incorporates information supplied by the manufacturer. Frequently, along with the powder, the supplier sends a material safety data sheet (MSDS) detailing the method of disposal and other pertinent information on the use of the powder and its potential danger to the user.

Once opened, sealed vials must be stored in a desiccator; some require refrigerated storage in a desiccator. Antimicrobials vary in their storage requirements. Aminoglycosides are stable at room temperature (in a desiccator), whereas β-lactams need to be kept at 20°C or lower. A general rule to heed is that the lower temperatures add to the risk of water condensation and the associated problems of ensuring adequate desiccation. When preparing to open a sealed desiccator unit, it is necessary to hold the unit and its contents at room temperature and permit both to equilibrate. Before being returned to the freezer or refrigerator, the desiccator should be resealed and the accumulated moisture desiccated. Because of their instability, some β-lactams (ampicillin, amoxicillin, and methicillin) should not be stored for longer than 6 weeks. Although there is a paucity of data on the stability of laboratory stock standards, Table 3.7 provides some relevant information (21).

When antibiotic stock solutions are being prepared, powdered (lyophilized) material must be completely dissolved. Because all antimicrobials are not soluble in water, Table 3.8 indicates the nonaqueous solvents that are usually used. Some solvents, such as dimethylformamide (DMF), are intrinsically antibacterial. Barry and Lasner (22) found that use of DMF as a solvent for nitrofurantoin introduced errors in nitrofurantoin diluted 1:50. However, if DMF was diluted in Mueller-Hinton broth (MHB) at 1:100 to 1:200 (i.e., 0.1 to 0.2 mL of DMF), there was no detectable inhibition or antagonism associated with the residual DMF concentration when tested against S. aureus or Escherichia coli (Fig. 3.1).

When stock solutions are prepared, they frequently need to be sterilized by filtration. At the level of concentrated stock solutions, it may not be necessary to filter-sterilize because these highly active solutions may be self-sterilizing. Aseptic precautions should be followed when further dilutions are warranted. When filtration through a matrix is needed, care should be taken to avoid the use of fiber pads owing to their absorbent nature, especially with antimicrobial agents of protein structure. The characteristics of four filter types in the retention of antimicrobial activity were studied by Murray and Niles (23) and are listed in Table 3.9.

Several factors influence the outcome and reproducibility of broth susceptibility results. First, the microorganism manifests its own genetic background, structure, and metabolic behavior, which strongly influence the development of resistance and disease-producing potential. Second, the antimicrobial agent possesses characteristics that affect solubility, protein binding, distribution, absorption, stability, and metabolic modification. The third factor is the milieu in which these interactions are tested and evaluated. These variables are noted in Table 3.2 and are discussed in the following pages. The problems associated with dilution schedules are discussed in “Concentration Range for Testing.”

Media

Ideally, the medium in which the microorganism and antimicrobial agent interact should mimic the serum or interstitial fluid of the patient. The problem of medium influencing the results and reproducibility of susceptibility tests has long been recognized. Liquid (and agar) media suffers from variations in the lots produced by the manufacturer, in part because of the variability of raw materials and the intrinsically undefined nature of the formulations. It is clear that no single medium could ever completely satisfy the goals of providing relevant reproducible end points for all antimicrobial agents tested against all potential pathogens. The requirements for an ideal medium (liquid or solid) have been summarized by Barry (24) as follows:

 1.  The medium should support the growth of a variety of pathogens for which susceptibility tests are required, without the need for special supplements or enrichments.

 2.  Medium contents should be defined, at least to the point of specific production details for crude components such as peptone and agar.

 3.  Different batches of the medium prepared by different manufacturers should yield reproducible results.

 4.  The medium should be free of components that are known to interact with antimicrobial agents that will be tested.

 5.  The medium should be capable of controlling pH (especially on the acid side) during the growth of common pathogens.

 6.  The broth and agar versions of the medium should have the same formulation, except for the solidifying agent.

 7.  The medium should be approximately isotonic for bacteria, and the agar version should be able to accept the addition of blood when required for the growth of fastidious microorganisms.

Over the years, several different media representing compromises from the ideal have been used. In a coordinated effort, the ICS (6) directed a study to determine the effects of variations in medium constituents. When two media, namely Grove and Randall no. 9 medium (25) and Mueller-Hinton medium (26,27), were compared (1), it was found by the ICS that Mueller-Hinton was more suitable for supporting the rapid growth of enteric gram-negative bacilli and group A streptococci, whereas Grove and Randall no. 9 produced better growth of these gram-positive organisms: Staphylococcus aureus, the enterococci, and Streptococcus viridans, the streptococci. When two commercial sources of Mueller-Hinton were evaluated, notable differences were found only for the aminoglycosides (at that time, streptomycin and kanamycin). It is likely that varying cation content accounted for these differences (discussed later).

MHB is probably the most widely recommended liquid medium for broth dilution tests (6,28–32). However, this medium was originally intended not for susceptibility testing but for the isolation of pathogenic Neisseria species (33). Because MHB contained low levels of paraaminobenzoic acid (p-ABA), it was used in agar form to determine the susceptibility of microorganisms to sulfonamide (34). The beef extract and casein hydrolysate in MHB are poor credentials for a reference medium. The former is difficult to standardize, and the high salt content of the latter needs to be compensated for. In spite of these deficiencies, the ICS group decided to use MHB in several studies because of the relatively good reproducibility and simplicity. An added feature was the low content of p-ABA, which made the medium suitable for testing sulfonamides. When MHB, brain-heart infusion (BHI), trypticase soy (TS), and Oxoid media were used in comparing tube (macrodilution) and broth microdilution results, MHB fared extremely well (35). In a replicate test series with S. aureus, none of the means of MICs varied by more than one dilution. For gram-negative organisms, the MICs were generally lower for MHB than for the other media tested. The greatest variation was with the Oxoid medium, which produced differences of as much as three dilutions between the macrodilution and microdilution methods. In another study in which the same four media were evaluated, Tilton et al. (36) found significant differences among the media when tested against reference strains of E. coli (Table 3.10), S. aureus, and Pseudomonas spp. The highest MICs were obtained in TS broth for the gram-negative strains. Differences between the media rarely exceeded 1.5 dilution steps, and the authors concluded that MHB, at cation-adjusted concentrations, was an acceptable medium. Other researchers have reached the same conclusion (37).

The issue of Mueller-Hinton being designated as a reference medium, even though its deficiencies have been recognized, has become moot. The medium has been used extensively, and there appears to be no proposals or contenders to replace it. CLSI (NCCLS) has established a reference lot for Mueller-Hinton agar (38). Cation-adjusted Mueller-Hinton broth (CAMHB), once recommended by CLSI (NCCLS) for routine testing of commonly encountered microorganisms, is now recommended when testing all species and antimicrobial agents. CAMHB is available from commercial manufacturers with Ca²⁺ (20 to 25 mg/L) and Mg²⁺ (10 to 12.5 mg/L) supplemented for convenience and uniformity. It is necessary to use CAMHB when aminoglycosides are tested against P. aeruginosa and when tetracycline is tested against other bacteria.

pH, Buffer, and Incubation

The mechanisms of the effect of antimicrobial pH are not precisely understood and are not consistent from drug to drug. In addition, the pH of the medium affects the activity of certain antibiotics. For some drugs, the pH variation is minor. For example, the nonionized side chain of penicillin G is slightly more active in acidic medium, but the effect is inconsequential, in that it can be demonstrated only by utilizing special experiments. The effect of the pH of the medium on the activity of six classes of antimicrobial agents is indicated in Table 3.11.

Penicillins that are nonionized or weak acids demonstrated minor variations according to the broth medium used, even when the pH was adjusted to approximately 7.0, and the MICs observed were generally within experimental error (one or two dilutions) for five different broths (39). This situation is in contrast to that of the aminoglycosides. Streptomycin is 500 times more active in alkaline medium than in acidic medium (40). Other aminoglycosides show similar but less drastic shifts in activity with changes in pH (Table 3.12).

The buffering capacity inherent in the formulation of a medium contributes to the stability of the pH during incubation; the inclusion of glucose in the medium results in some lowering of pH during the growth of strains that can ferment the substrate. These differences were observed in medium with and without glucose when E. coli was used as the test organism (24). The effect of adding specific buffers to stabilize a defined synthetic amino acid medium (SAAM) (41) has also been demonstrated.

Does the pH of the medium modulate the outcome of susceptibility results to the extent that the interpretation of the test and its clinical relevance would be altered? Can the pH environment of the patient be clinically manipulated significantly for therapeutic purposes? The answer to both questions is in the affirmative. Such examples probably occur more readily with the macrolide and aminoglycoside groups, which are more active in slightly basic media, and the tetracyclines, which are more active in acidic environments. Definitive changes in interpretive results have occurred with gentamicin, kanamycin, erythromycin, novobiocin, and tetracycline (15) (Table 3.13).

The marked pH effect on the aminoglycosides may be explained by the degree of ionization of these compounds. However, this explanation is not plausible for changes with the prototypic penicillin G, which is intrinsically slightly acidic but allows for minimal changes of antibacterial activity in the range pH 6 to 8. Another mechanism proposed for nonionized erythromycin may be directly responsible for antimicrobial activity (42). The antimicrobial activity of this macrolide is enhanced vividly as the pH approaches a pK_a of 8.6 (Fig. 3.2). This enhanced activity of erythromycin at alkaline pH extends the spectrum of activity of this compound to include gram-negative organisms, particularly E. coli (43). These observations have been used in the clinical treatment of urinary tract infections. It has been shown that alkalinization of the urine is sufficient to activate erythromycin and produces successful clinical treatment (42). Theoretically, laboratories should set up an environment that mimics this pH difference, but rarely is this needed or accomplished.

Another example of the effect of pH on the outcome of susceptibility test results and treatment can be found in studies with Helicobacter species. During the past several years, Helicobacter organisms have been implicated in a wide variety of conditions involving the stomach and duodenum, from gastritis to dyspepsia, as well as potentially gastric carcinoma (44). Helicobacter pylori was formerly referred to as Campylobacter pylori and is generally regarded as the agent associated with or implicated in these conditions. H. pylori is susceptible to a wide variety of antimicrobial agents, including penicillins, cephalosporins, macrolides, nitrofurantoins, and quinolones (44,45). The relative efficacy of these treatments is unknown, and many patients have gastric intolerance as a side effect (44). It is uncertain at which pH antimicrobial agents demonstrate maximum effectiveness in vivo. Similarly, it is questionable at what pH antimicrobial susceptibility testing should be executed. Recently, Grayson et al. (46) tested the susceptibility of 22 isolates of H. pylori (obtained from gastric mucosal biopsies) against eight antimicrobial agents. Antimicrobial susceptibility studies utilized the agar dilution methodology with an inoculum of 5 × 10⁵ CFU/spot. The comparative efficacies of the agents are shown in Table 3.14. The macrolides, quinolones, and clindamycin demonstrated diminished activity at acidic pH, compared with their activity at pH 7.4. In the alkaline pH range, MIC₉₀ values were moderately but uniformly decreased (increased potency) for these same drugs. The MICs for ampicillin and metronidazole were essentially constant and within clinically achievable levels in the three pH ranges tested. The authors concluded, based on the results of their study, that if the pH of a human gastric environment was only slightly below pH 7.4, a significant loss of antimicrobial activity could result. They suggested that the modulating effect of pH on antimicrobial activity may explain the apparent discrepancy between in vitro susceptibility results and observed therapeutic failures and that clinicians need to consider the effect of pH when selecting an agent for the treatment of H. pylori. Added data from a study of 30 clinical isolates of H. pyloriwith trospectomycin, ampicillin, metronidazole, clarithromycin, azithromycin, and clindamycin under varying pH conditions showed that an acidic environment unfavorably affected the activity of all of the agents tested (47).

The gaseous environment is important in stabilizing the pH of media, particularly solid media; it can also affect broth. The presence of carbon dioxide (CO₂) in the incubation atmosphere, for example, may produce acidic changes, at least early in the course of incubation. Standardized susceptibility testing should not be performed in a CO₂ environment. If capnophilic organisms are being tested, their growth would be scanty in an aerobic atmosphere, and so they can be incubated in CO₂, but interpretation of the results is not considered routine. When susceptibility results obtained in an aerobic environment and a 10% CO₂ incubator were compared, the MICs of aminoglycosides and erythromycins were higher (i.e., the antimicrobials were less active) and tetracyclines, methicillin, and novobiocin were more active in the CO₂ environment (6). These studies pertained to diffusion tests; however, the net effect of the 10% CO₂ environment was a lowering of pH.

Although one would expect that the target temperature for incubation of antimicrobial susceptibility tests would be between 35°C and 37°C, the temperature of incubation has been studied, particularly in regard to the detection of MRSA. Because MRSA strains are known to grow at reduced rates, it would be anticipated that these would be favored at lower incubation temperatures, compared with their more rapidly growing susceptible counterparts. In a study by Mackowiak (48), incubation at 35°C proved satisfactory for routine growth of MRSA strains as well as for all susceptibility tests. However, the heteroresistance of MRSA may be missed at 37°C. At elevated temperatures (from 35°C to 41.5°C), incubation of routine susceptibility tests (microtiter and serum bactericidal tests) indicated that more than 20% of the MIC values were lowered significantly (Fig. 3.3), and serum activity was enhanced at the higher temperatures.

The molarity (i.e., the strength) of a buffer used in the formulation of a medium can also affect the antimicrobial activity of certain compounds. In tobramycin assay systems, it was shown that the activity of the drug was generally increased with decreasing molarity, and the effect was dependent on the type of assay system used (49). For gentamicin, increasing concentrations of phosphate buffer in concert with shifts in pH produced a greater degree of error.

The extent of incubation has been studied in relationship to its effect on MIC. As incubation was prolonged, increasing MICs of cephalothin were observed beyond 12 hours (50). Such effects may be partly due to progressive antibiotic inactivation with increasing incubation interval and are more likely to be encountered with the relatively unstable penicillins and cephalosporins (51,52). Broth and agar dilution methods were compared (53) in relationship to the duration of incubation (12 and 24 hours) (Table 3.15).

For the broth dilution techniques, the MICs increased at least twofold more than twice as often with prolonged incubation, compared with the changes that occurred between 12 and 24 with agar dilutions. This effect can be observed for ampicillin and cephalothin as well as newer broad-spectrum β-lactams. In broth dilution systems, the best reproducibility and minimal alteration in MICs from the late growth of residual persister cells can be seen if a routine is followed of reading broth dilution end points after overnight incubation (usually 18 hours). (See “Automated, Rapid, and Instrument-Associated Methods” for the adverse effects of short-term incubation.)

When microorganisms are exposed in broth to antibiotic concentrations for short durations and then the antibiotic is removed or inactivated, the resultant effect has been termed the postantibiotic effect, defined by Craig and Gudmundsson (54). This phenomenon may be relevant to clinical responses to therapy and can be used to establish dosage schedules (55,56). Extended incubation over a number of hours in antibiotic-deficient medium must be used to determine the residual effect of brief exposure (0.5 to 4 hours) to the antimicrobial agent, so this type of assay cannot be considered a routine or rapid test.

Cation Concentration and Osmolality

The divalent cations calcium and magnesium have a profound modulating impact on the effects of the aminoglycosides, especially gentamicin. For example, the general results of broth dilution tests with gentamicin show MICs 1 log₂ unit higher than the results for agar medium, probably reflecting variations in cation concentrations. Other aminoglycosides, as well as tetracycline, are also affected. Varying cation content has posed significant problems in agar medium (57). However, different concentrations of magnesium in broth medium have also been responsible for marked MIC variations in the activity of gentamicin against many Pseudomonas strains (58). Falsely low MICs are observed with media not supplemented with calcium and magnesium. A similar but less pronounced effect can be seen against other gram-negative organisms in unsupplemented media (58,59). The data in Table 3.16 indicate results obtained by varying the concentrations of magnesium, calcium, and sodium salts in nutrient broth for susceptibility studies of Pseudomonas with gentamicin. After extensive surveys (60), it was suggested that final concentrations of magnesium of 20 to 35 mg/L and concentrations of calcium of 50 to 100 mg/L should supplement broth. These concentrations modify the medium and bring it into the physiologic concentration range found in serum (Table 3.17). In commercial microdilution systems, the medium is cation-supplemented.

In studies with the tetracyclines, different MICs were obtained in TS broth compared with MHB (62). These differences were attributed to variations in cation concentrations. MHB produced eightfold lower MICs against S. aureus, E. coli, and Klebsiella species, but the antibiotic inhibitory effect of the TS broth was reversed by the apparent chelating effects of phosphate, oxalate, or citrate. The opposite effects (increased MICs) on tetracycline in MHB were produced by adding the cations magnesium, calcium, or iron (62).

Sodium chloride (NaCl) concentrations change the osmolarity of medium and have a marked effect on the activity of the aminoglycosides gentamicin (63) and tobramycin (49). A sixfold decrease in activity and increase in MIC was noted by changing (increasing) the NaCl concentration from 22 to 174 mmol/L.

β-Lactams that bind to varying penicillin-binding proteins (PBPs) may be uniquely affected by the osmolarity and, in turn, the conductivity of the test medium. This effect of osmolarity varies from strain to strain of bacterium. For some strains of Proteus or Klebsiella tested against mecillinam, a 500-fold increase in MIC as a result of increasing molarity from 185 to 402 mOsmol/L was observed. In contrast, a strain of Enterobacter cloacae showed no change in the MIC for mecillinam between the same osmolarity levels in the same broth (NIH medium) (64).

Supplements and Other Additives

Recently, MHB has been shown to be acceptable for a wide variety of antimicrobial susceptibility tests and to be well suited for standard, rapidly growing pathogens such as enteric gram-negative bacilli, Pseudomonas spp, staphylococci, and Enterococcus spp. For organisms that grow rapidly and have been studied adequately, there are well-standardized interpretive guidelines and QC standards (10,11). For bacteria that do not grow readily on this medium, other supplements or alternative media may be required. The addition of blood or hemoglobin or of special additives such as IsoVitaleX (Baltimore Biological Laboratories, Baltimore, MD) has been proposed to enhance the growth of many fastidious organisms that may frequently require susceptibility testing (see “Fastidious and Unusual Pathogens”). Because some of the supplements produce an opacity that cannot be readily used in standard macrodilution or microdilution broth systems, they are more suitable to solid agar medium. The susceptibility testing of H. influenzae has long been a problem in this regard. Only recently, Jorgensen et al. (65) studied the problem and developed a special test medium, Haemophilus test medium (HTM), which has become commercially available and provides a solution to this problem.

The addition of various supplements of unknown chemical composition may in some instances alter antibiotic activity. However, if adequate controls are incorporated, standards are developed for appropriate interpretation, and additional control strains are utilized in the testing protocol, media supplemented with various components may be used successfully for broth dilution susceptibility tests.

Supplementation of test media with previously used (but undefined) components can result in unusual effects. Eliopoulos et al. (66) reported on the effect of 5% sheep blood added to Mueller-Hinton agar on the activity of cefotaxime and other cephalosporins against Enterococcus faecalis. They determined that the activities of cefotaxime and other aminothiazoyl oxime cephalosporins (e.g., cefpirome, ceftazidime, cefmenoxime, ceftriaxone, and ceftizoxime) against E. faecalis were enhanced by the addition of 5% sheep blood. This effect was not documented for aztreonam (a nonoxime aminothiazoyl), cefotiam, or other cephalosporins and was specific to the syn-configuration of the oxime moiety. Enhancement of cefotaxime activity was shown against 50% of 85 clinical isolates and could be demonstrated only with low bacterial inocula. The α-globulin fraction of serum mimicked this enhancing activity, whereas α₁-, β-, and γ-globulin fractions and albumin frequently antagonized or did not significantly affect the antimicrobial activity. Although these observations may not have direct clinical relevance, they present a possible explanation for the relatively infrequent occurrence of enterococcal superinfection in patients treated with cefotaxime, which demonstrates poor in vitro activity.

When 5% sheep blood or 10% fetal calf serum was added to liquid media for testing H. pylori susceptibility (67), sheep blood inhibited growth. When the effect of bismuth was evaluated, it was found that the compound inhibited growth in medium containing starch but that the inhibition was neutralized in medium containing serum (Fig. 3.4).

Prior to the development of HTM, broth dilution tests were satisfactorily done using MHB (68). Today, susceptibility testing of sulfonamides does not pose a serious, relevant clinical problem. Sulfonamides are rarely used in treatment, and their testing is not usually required. Media enriched with various digests or supplements may contain p-ABA, thus obviating sulfonamide inhibition and rendering the tests inaccurate or useless. Testing of trimethoprim or a combination of trimethoprim and sulfamethoxazole presents similar problems, in that the activity of trimethoprim or sulfamethoxazole is antagonized by the high thymidine content of enriched media. This problem exists in vitro although apparently not in vivo, because thymidine and thymine both seem to be present in sufficiently low levels in blood and urine so as not to interfere with in vivo bacteriostatic and bactericidal activity during treatment (69,70). Thus, media have to be adjusted so that the thymidine content is diminished or absent. A suitable formulation was designated SF medium base (71). Dramatic differences between a standard broth medium and one enriched with 5% lysed horse blood can be seen in Figure 3.5. The supplement of 5% lysed horse blood contains sufficient thymidine phosphorylase to inactivate thymidine in various media (71,72). In the United States, most commercial lots of Mueller-Hinton medium have proven satisfactory for overnight incubation as a result of careful QC measures by the manufacturers. Other technical problems encountered with broth dilution methods when testing sulfonamides and trimethoprim result in hazy end points. These problems have been resolved by manipulation of the medium and the use of small inocula (72).

Other Media

It appears that MHB and CAMHB have established a foothold in the antimicrobial susceptibility literature. Presently, no other medium has been studied sufficiently to replace them. There have been investigations into one defined medium, namely, SAAM (41). Although this medium proved comparable in growth support and antagonism of sulfatrimethoprim, it has not gained acceptance for routine use. Similarly, another defined medium free of purines or pyrimidines (73) has been studied. However, a large proportion of Streptococcus and Staphylococcus strains failed to grow on this medium; thus, it is unsuitable for use in routine clinical laboratories. There has been little investigation into other media that would meet the ideal criteria defined by Barry (88,89). It is more likely that, with the aid of organizations like CLSI (NCCLS), reference lot media may be designed and accepted. As noted, a reference lot has been defined for Mueller-Hinton agar with regard to susceptibility testing of H. influenzaeon agar surfaces.

The density of inoculum in an antimicrobial susceptibility assay is critical for the generation of reliable and reproducible susceptibility test results. The adjustment of inoculum density is more important for broth dilution tests than for disk diffusion and agar dilution methods. The reason is that, with agar methods, visual (macroscopic) inspection of growth permits semiquantitative evaluation of inoculum density. In contrast, it is difficult to estimate the growth density in the growth control tubes (Table 3.1).

Susceptibility test results differ according to the species, strain, and antimicrobial agent tested (36,53,74,75). During growth of a particular bacterial strain, it can be assumed that homogeneous progeny develop. However, some degree of inoculum size effect may result from the MICs of individual bacterial cells that follow a normal distribution. Thus, with larger inocula, there is a greater probability that there will be some cells or variants from the more resistant end of the distribution curve. In broth, the cells from this extreme of the normal distribution are more likely to survive and grow.

As mentioned, large bacterial populations are less promptly and completely inhibited than smaller ones. Also, the likelihood of the emergence of resistant mutants is greater in a large population of bacterial cells. This is exemplified by the emergence and recognition of MRSA and methicillin-resistant Staphylococcus epidermidis (76). To detect and define these particular populations more accurately, it is necessary to depress the growth of the more rapidly growing, susceptible population and to extend the incubation interval beyond the usual 18 to 24 hours so as to readily detect the presence of the more slowly growing, resistant population (77). There is less chance of observing the resistant population if the critical inoculum is small. According to Sanders et al. (78–80), low-frequency mutant subpopulations of antimicrobial-resistant organisms are best detected by broth dilution systems at an inoculum concentration of more than 10⁵ CFU/mL.

The inoculum effect, first reported in 1940 for the interaction of Streptococcus haemolyticus and a sulfa compound (81), was later described for penicillin and S. aureus (273). The effect has been found in several bacterial species and is particularly widespread among the β-lactam antimicrobial agents when their activity is directed against β-lactamase–producing bacteria. Although the inoculum effect has been most widely studied in the staphylococci, it has been shown to be associated with a variety of bacterial species and almost every class of antimicrobial agent. Table 3.18 outlines the antibiotic–organism pairs that generally exhibit an inoculum effect. This effect is generally attributed to the inactivation of the antimicrobial agent by β-lactamase (82). However, it is known to occur with antimicrobial agents lacking the β-lactam ring. Other possible causes of the inoculum effect include the selection of resistant mutants and drug breakdown by other drug-targeted inactivating enzymes. The inoculum effect can be defined as a significant increase in MIC (plus two dilutions) when the inoculum size is increased (at least by 0.5 log unit). In some studies, the effect was noted when the inoculum was varied by four orders of magnitude. Early studies investigating this phenomenon need to be considered in light of the standardized elements of testing recommended by CLSI (NCCLS) (10). Additionally, studies elucidating the inoculum phenomenon have not been well standardized; they vary in incubation interval, reagent volume, and size of the test vessel.

The clinical implications of the inoculum effect are uncertain. Clearly, the inoculum standard established by CLSI (NCCLS) (5 × 10⁵ CFU/mL final concentration for broth dilution and 10⁴ CFU/spot) is not applicable to all clinical situations. At best, it represents a compromise between various clinical infections and the procedural manipulation for eliminating the potential for trailing end points. This methodologic result occurs when the inoculum exceeds a final concentration of 10⁷ CFU/mL. With several of the antibiotic–organism pairs that have exhibited the inoculum effect (Table 3.18), the final MIC result, although elevated compared with the result obtained with a smaller inoculum size, would still have yielded an interpretation of susceptible based on the MIC:level ratio.

Of the several factors that can modulate the outcome of antimicrobial susceptibility tests, the inoculum size, besides being relatively easy to measure, should be among the easiest to standardize. The role of inoculum size was shown in studies of penicillin-sensitive and penicillin-resistant strains of S. aureus treated with cephalexin (83,84). Marked changes in MICs and MBCs were observed for 100-fold changes in inoculum density (Fig. 3.6). The effects of 10-fold dilutions of inoculum, from 10³ to 10⁷ CFU/mL, can be seen for several antibiotic groups used against E. coli (Table 3.19), gram-positive and gram-negative organisms (Table 3.20), and Pseudomonas spp (Tables 3.21 and 3.22).Against E. coli, tetracycline was found to be the most refractory to inoculum effects, whereas significant changes were recorded with the other antimicrobial agents, especially at the high end of the inoculum range (Table 3.10). For P. aeruginosa, the effect of varying inoculum size on the activity of the antimicrobials tested was related to the agent tested (75). When the inoculum was increased to 5 × 10⁵ CFU/mL, the MIC₉₀ values for all drugs tested were increased (Table 3.21).When the inoculum was increased further to 5 × 10⁷ CFU/mL, the MIC₉₀ values could be determined only for gentamicin and thienamycin. The MBC₉₀ values at an inoculum of 5 × 10⁵ CFU/mL ranged from 8 µg/mL for gentamicin and thienamycin to 128 µg/mL for cefotaxime (Table 3.21). With the largest inoculum, the MBC₉₀ values for gentamicin and thienamycin remained constant but the MBC₉₀ values for the other drugs tested were less than 128 µg/mL. The susceptibility results for H. influenzae are seriously influenced by the size of the inoculum (85–87). The effect is more pronounced for ampicillin-resistant isolates (β-lactamase producers?) and penicillin than for cephalosporins. Now that inoculum size and medium have been defined for H. influenzae susceptibility testing, variation in susceptibility test results is anticipated to diminish.

The effect of inoculum density variation on various antimicrobial agents tested against several species must be considered in any recommendation for a standard or reference method. Most investigations support the earlier work of the ICS (6), which suggested that an inoculum of 10⁵ to 10⁶ CFU/mL would yield acceptable results in a macrobroth dilution test. Some workers adjust the inoculum closer to the range of 1 to 5 × 10⁵ CFU/mL. For microdilution, the average recommended inoculum is 1 × 10⁶ CFU (30).

Methodology for Standardizing Inocula

For the reasons mentioned earlier, it is imperative that each culture inoculum be individually standardized. The procedures frequently used for both the macrodilution and microdilution systems involve either adjustment of a logarithmic-phase broth culture to a McFarland 0.5 turbidity standard (7,10) or defined direct dilutions from 0.5-mL volumes of stationary-phase broth cultures (10). Several discrete colonies, usually three to seven, are subcultured to the inoculum growth broth, to avoid single-colony variance. The inoculum is usually cultured in the same broth medium used for the test, such as MHB or CAMHB (richer media such as TS broth and BHI can also be used satisfactorily). For rapidly growing pathogens, overnight broth cultures (4 to 8 mL) grow to approximately 10⁹ CFU/mL. A 1:1,000 or 1:2,000 dilution of this growth brings the inoculum density to the range of 5 × 10⁵ to 5 × 10⁶CFU/mL. Results have generally been satisfactory with such methods, because most rapidly growing pathogens achieve stationary-phase growth at density levels that fall within a reasonable range. Alternatively, one can adjust the turbidity of a 4- to 8-mL overnight culture or a 4- to 6-hour broth culture (both are considered to be in the stationary phase) to a standardized density. This can be accomplished nephelometrically with instruments frequently provided with automated units for determining susceptibility or manually by dilution to match the visual turbidity of the McFarland 0.5 BaSO₄ standard, which approximates 10⁵ CFU/mL. The 4- to 6-hour broth culture offers a significant time advantage compared with the overnight incubation. However, for some more slowly growing, more fastidious strains, such as Haemophilus or Neisseria strains, overnight growth may be required. Satisfactory results have been obtained by emulsifying colony plate growth to match either nephelometric or McFarland turbidity standards. Then, appropriate dilutions (e.g., 1:2,000) are made for the final inoculum.

It has been shown that turbidity-adjusted direct suspensions can also yield reproducible and accurate results for commonly encountered gram-negative or gram-positive, rapidly growing microorganisms (44). When the growth phase of the inoculum was studied systematically by Barry et al. (88), they found that generally satisfactory reproducibility with control strains was dependent on whether the inoculum standard utilized required harvesting direct suspensions from overnight colonies or from logarithmic-phase (2- to 4-hour) or stationary-phase (5- to 6-hour) broth cultures.

After an inoculum has been prepared, variability has been observed in the delivery of the inoculum to the test system. In some systems, this was more dependent on a mechanical inoculum preparation and transfer system and independent of the growth phase of the organism. Inocula prepared with Pseudomonas strains demonstrated lower counts, perhaps due to clumping, than did those prepared with other organisms (89). Final MIC results were generally reproducible and consistent regardless of the method of inoculum preparation and despite the system-dependent inoculum variations from 2 × 10⁴ to 1 × 10⁶CFU/mL (44). Table 3.19 indicates that, with some drug–organism combinations (10%), statistically significant differences in geometric mean MICs resulted from different growth or inoculum preparation methods tested in two different microdilution systems (89).

Numerous techniques have been developed for inoculum preparation, standardization, and transfer and been incorporated into the expanding range of commercial microdilution test systems (see “Standard Broth Dilution Procedures”). Several systems adjust turbidity suspensions instrumentally. However, most recommend the use of a mechanical device that involves touching five colonies directly with a wand containing grooves at its bottom crosshatch, followed by emulsification in an appropriate volume of saline. This system, referred to as RISS and marketed as PROMPT, eliminates prior broth incubation and turbidity adjustment (90,91). Figure 3.7 illustrates how the CFU/mL values in the inoculum are affected by increasing the number of colonies touched; a maximum colony count is achieved after selecting four colonies.

The focus of this chapter is on broth dilution procedures. However, it is worthwhile to note that comparisons of results of broth and agar dilution methodologies have shown them to be similar (92). In some instances, as noted by Lorian (93), the differences for quinolones have been significant. There has been a considerable range of variability in studies, depending on the method used, the antimicrobial agent studied, and the organisms tested. A number of investigations have also compared microdilution and macrotube dilution techniques (37,94,95). Tables 3.23 and 3.24 present some of the results. Clearly, for the majority of agents tested, there were few strains that varied by more than plus or minus one dilution.

Comparative evaluations of the microdilution and macrodilution broth procedures are noted here to provide a historical perspective. Since the 1980s, there have been no evaluations of these two broth methodologies, owing in part to the popularity of the microdilution procedure. Furthermore, commercial applications of the miniaturized technique that began in the mid-1970s have made it practical for laboratories to provide MICs for all isolates, or at least for all those they wish to report on. In general, microdilution test results are essentially equivalent to those obtained with the standardized macrobroth dilution procedure. For certain organisms, such as gram-negative bacilli, the microdilution results are approximately one dilution step lower than those obtained with the macrodilution broth technique. This seems to be a general feature of the microdilution technique and may be partially due to the way end points are read. Minimal turbidity, which is visible in a test tube, may not be readily observed in a microtiter well. Another possible explanation may involve the inoculum. Although in their final concentrations the inocula in the two systems contain approximately 5 × 10⁵ organisms and are comparable, the absolute number of viable cells delivered into the microtiter well is 1 log unit less than that inoculated into the test tube.

Standard Microdilution Broth Procedure

The performance of microdilution tests has been described by a number of authors using in-house panels prepared by mechanical semiautomated or automated equipment (29,94,96–101) or commercial test panels routinely prepared or customized for the user. Commercially prepared test panel systems employ either frozen antimicrobial solutions in medium or dried preparations that need to be rehydrated by adding diluent and inoculum to each well. The test panel state of some commercial preparations and their method of inoculation are noted in Table 3.25.

For preparation of standard microdilution panels in the laboratory, the following general recommendations and directions can be followed (29,102). The antimicrobial agent stock solutions are prepared and stored as noted in Tables 3.4 and 3.7 and as found in CLSI (NCCLS) standard M7-A6 (10). The recommended dilutions have been included in Table 3.4, and the flexibility dilution schedule need not be limited to serial twofold increments.

The recommended broth for testing is Mueller-Hinton supplemented with 50 mg/L Ca²⁺ and 25 mg/L Mg²⁺. These concentrations are achieved by the addition of the appropriate amounts of filter-sterilized CaCl₂ and MgCl₂ stock solutions to cold sterile broth. Ideally, microdilution trays should be prepared each day they are used. However, with the availability of semiautomated dispensing devices for preparing dilutions and dispensing prediluted agent, it is possible to prepare large numbers of trays at one time and store them frozen until required. As trays are filled, they are stacked in groups of 5 to 10 and are either covered with an empty tray on top or placed and sealed in plastic bags and frozen at 20°C or 70°C. Household freezers are satisfactory, but they should not contain self-defrosting units, because fluctuations in temperature during the defrost cycle thaw and refreeze the antimicrobial agents and thus contribute to their deterioration. Trays quick-frozen at 70°C and stored at 20°C have a useful shelf life of about 6 weeks. Storage at 70°C significantly increases the shelf life to approximately 3 months. Once a group of trays is removed from the freezer, they should be allowed to warm at room temperature prior to use. Unused thawed trays should be discarded and never refrozen.

Actively growing broth cultures are diluted to a McFarland value of 0.5 (as described earlier). Multipoint plastic or metal replicators (inoculators) are used in several commercial and semiautomated systems. After inoculation, each well should contain approximately 5 × 10⁵ CFU/mL (5 × 10⁴ CFU/well).

After inoculation of microdilution trays, they should be covered with sealing tape to minimize evaporation. Alternatively, large numbers of trays can be stacked and covered with an empty tray. After 16 to 20 hours of incubation, trays may be examined from below with a reflective viewer and MICs determined. The trays can also be read visually from the top. The end point MIC is the lowest concentration of drug at which the microorganism tested does not demonstrate visible growth. In judging the end point, it is necessary to compare the growth (or absence of growth) in the test wells with the growth (or absence of growth) in the well without the antimicrobial agent. End points are easily read as turbid wells or clear wells. For some antimicrobials, such as the sulfonamides and trimethoprim, end points may trail. The end point for these drugs should be read as an 80% to 90% decrease in growth compared with growth in the control well.

Standard Macrodilution Broth Procedure

The variables that affect the outcome of dilution tests were outlined earlier and were reviewed in the preceding section. The procedures for performing the macrodilution broth test have been described by several authors (6,24,25,32,103–106). Errors and statistical variability of dilution procedures have been reviewed (30). The basic methodology described in the ICS report (6) has proven satisfactory for many laboratories, and Table 3.5 presents a scheme recommended in the ICS report for the preparation of antibiotic dilutions. An alternative dilution schedule for preparing a larger number of tubes is shown in Table 3.6. Because the dilutions are pipetted directly into blocks of three dilution tubes with one pipette, the chance of error is significantly reduced. A sequential twofold dilution method is summarized from the work of Jones et al. (107) as follows. The working antimicrobial solution is prepared by diluting the drug and MHB to the highest final concentration desired. The test is performed in 13 100-mm screw-capped (or cotton-plugged, etc.) test tubes. For a limited number of tests, twofold dilutions are prepared directly in test tubes as follows: 2 mL of the working solution of the drug is added to test tube 1 of the dilution series. To each remaining tube, 1 mL of MHB is added. With a sterile pipette, 1 mL is transferred from tube 1 to tube 2. After thorough mixing, 1 mL is transferred (using a separate pipette for this and each succeeding transfer) to tube 3. This process is continued to the next to last tube, from which 1 mL is removed and discarded. The last tube receives no antimicrobial agent and serves as a growth control. The final concentrations of antimicrobial agents in this test are half those of the initial dilution series because of the addition of an equal volume of inoculum in broth. The inoculum is prepared and adjusted, as noted earlier, to contain 10⁵ to 10⁶ CFU/mL, by adjusting the turbidity of the broth culture to match the McFarland 0.5 standard. It is then further diluted 1:200 in broth, and 1 mL of the adjusted inoculum is added to each test tube. Tubes are incubated at 35°C for 16 to 20 hours. The lowest concentration of antimicrobial agent that results in complete inhibition of visible growth represents the MIC. A very faint haziness or a small button is usually disregarded.

Quality Control

For in vitro susceptibility test results to be meaningful for selecting appropriate antimicrobial agents and monitoring their use for the treatment of infection, they need to be accurate and reproducible. Because of the potential for variation, emphasis has been focused on strict adherence to methods and reference procedures (10,11). The development of QC parameters has played a significant role in the high level of performance obtained by most laboratories. Keys to this performance are the application of standard reference strains with known reactivity and the assessment of qualitative and quantitative end points.

The ideal reference strain for QC of dilution susceptibility methods should have MIC end points near the middle of the range of concentrations being tested for a given drug, or at least no closer than two dilutions from the extremes of the test range included (102). Thus, in QC for dilution testing, it has been necessary to deviate from established QC strains that have been used for disk diffusion testing. For example, S. aureus American Type Culture Collection (ATCC) strain 29213, a weak β-lactamase producer, is recommended instead of ATCC strain 25923. Additionally, S. faecalis ATCC strain 29212 and E. coli ATCC strain 35218 have been recommended as controls for β-lactamase inhibitors such as clavulanic acid and sulbactam. A sixth control strain, H. influenzae ATCC strain 49247, has recently been proposed (by CLSI [NCCLS]) for testing drugs against H. influenzae (10).

Studies of the precision and accuracy of macrodilution and microdilution MIC end points with the older established QC strains have shown that reproducibility should be within plus or minus 1 log₂ dilution interval for 95% of the replicates (32,103) and that most tests should fall at the modes. The ranges of MICs expected for several contemporary antimicrobial agents against the reference strains are indicated in Table 3.26.

Control reference strains should remain genetically and phenotypically stable over many replications and long-term storage. Control strains must be stored using procedures designed to minimize chances of mutation or variant selection (32). It is suggested that, for long-term storage, QC strains be lyophilized or frozen in a stabilizing medium such as whole sheep blood or 15% glycerol in an enriched broth such as BHI or 50% serum in broth. Freezing at 60°C (or below) is preferable to storing in conventional freezers (20°C). Strains can be maintained for short-term storage at 4°C for approximately 2 weeks on agar (soybean digest casein).

When new batches or lots of microdilution trays are received from a commercial source or prepared, they should be tested with reference strains to determine their acceptability. The MICs resulting from QC testing should be no more than one dilution interval above or below the anticipated MIC. If the difference is greater, either the batch is rejected or the results with the affected antimicrobial agent are not recorded or reported. Additionally, representative uninoculated trays should be tested for sterility of the medium. QC should be done on a periodic basis after reproducibility and accuracy have been documented by daily QC practice (10,108). These QC procedures provide a review of variables such as antimicrobial potency and stability, instrument function, and technical proficiency. The MICs obtained with each reference strain should be maintained in a record book for ongoing review.

Some additional QC procedures are necessary for broth dilution tests but are less critical for agar or disk diffusion tests. A purity control plate for each isolate tested is subcultured, for isolation on an appropriate agar medium, directly from the final inoculum suspension to detect potential contamination or mixed cultures. A growth control tube (or well) free of antimicrobial agents is included within each panel set to ensure adequate growth. This also serves as a turbidity control (a comparative aid) when reading end points. The inoculum should be measured periodically by a direct dilution plate count of the time 0 inoculum from the growth tube or well. Finally, the proficiency of the observers determining end points should be monitored periodically by comparing their results with those from a standard reader to ensure uniformity between different observers reading the same plate.

There are several types of susceptibility test results that, when obtained, require further confirmation or investigation. Forty-five nonsusceptible phenotypes are listed in Table 3.27, and these may be associated with pre- or postanalytical (test) errors. In any event, these phenotypes warrant follow-up study.

The standard broth procedures described thus far are applicable to routine antimicrobial susceptibility testing. The term routine reflects the fact that the testing is of rapidly growing, nonfastidious pathogens frequently encountered in the clinical setting. Routine susceptibility procedures, however, may not be applicable to predictably slow-growing microorganisms with prolonged lag times and/or slow generation times (109). Some clinically significant bacteria have characteristics that preclude their being tested by standard methods. They may grow too slowly, may require special nutrients or atmospheres, or simply may not have been tested with enough frequency to demonstrate that they can be tested accurately and reproducibly by the standard methods. These microorganisms have been termed fastidious and/or unusual. Specifically, fastidious organisms do not readily grow on Mueller-Hinton medium without supplementation. Unusual organisms may grow well on Mueller-Hinton medium, but studies have not been completed to demonstrate that they can be tested reliably by standard methods. Testing the susceptibility of unusual organisms to antimicrobial agents may also present special problems.

In the past, many of these organisms did not require susceptibility tests because they were known to be universally susceptible to an appropriate antimicrobial agent that was not toxic and could reach sufficient levels to effect a clinical cure. “Resistant” strains have emerged, however. Resistance to β-lactams is most often due to production of β-lactamase, which may be constitutive or inducible and may be mediated by either chromosomal or plasmid genes (80,110). In some species, such as H. influenzae and N. gonorrhoeae, penicillin and ampicillin resistance has been typically caused by the acquisition of plasmids that mediate constitutive β-lactamase production by the organisms (111,112). In other species, such as S. pneumoniae, penicillin resistance is not due to β-lactamase production caused by plasmids but is chromosomally mediated and is due to the alteration of PBPs (113).

These developments have produced challenges for clinicians because previously employed empirical therapies associated with these fastidious or unusual organisms may not be adequate. Additional laboratory testing is sometimes required to support a prediction of therapeutic success.

For some organisms, such as group A streptococci, susceptibility tests are not necessary because these organisms have maintained (with some exceptions) universal susceptibility to penicillin, the drug of choice. However, for S. pneumoniae, especially if it was isolated from a putatively sterile site, a test to determine susceptibility to penicillin is indicated, because some of the strains may be relatively or completely resistant to penicillin. If it becomes necessary to perform susceptibility tests on clinical isolates for which no standard method has been described, it is usually best to determine the MIC using the general broth dilution method described earlier and in CLSI (NCCLS) standard M7-A6. For testing of infrequently isolated or fastidious, see CLSI document M45-A2. Table 3.28 details the antimicrobial susceptibility methods that can be used for these fastidious or unusual organisms. It is obvious that certain microorganisms have been omitted, notably anaerobes, Mycobacterium tuberculosis, Chlamydia spp, Mycoplasma spp, and spirochetes. For these groups, readers can refer to appropriate chapters in this volume.

For the antimicrobial susceptibility testing of fastidious and/or infrequently encountered bacteria, the CLSI has published guidelines for testing with interpretive breakpoints (114). These guidelines encompass the following bacterial groups.

Aeromonas spp

Bacillus spp (other than Bacillus anthracis)

Campylobacter coli and Campylobacter jejuni

Corynebacterium spp

Erysipelothrix rhusiopathiae

HACEK group (Aggregatibacter actinomycetemcomitans and Aggregatibacter aphrophilus, Cardiobacterium hominis, Eikenella corrodens, and Kingella kingae)

Lactobacillus spp

Leuconostoc spp

Listeria monocytogenes

Moraxella catarrhalis

Pasteurella spp

Pediococcus spp

Plesiomonas shigelloides

Streptococci, nutritional dependent spp (now Abiotrophia and Granulicatellas spp)

In addition to the microorganisms listed earlier, the CLSI has proposed guidelines for the standardized testing of the generally recognized agents of bioterrorism, namely Bacillus anthracis, Francisella tularensis, Brucella spp, Yersinia pestis, and Burkholderia pseudomallei. (See Table 2 of CLSI M45-A2 [114].) As of this writing, standardized testing for other fastidious bacteria, for example, Legionella and Bordetella, are unavailable because these less frequently encountered bacterial infections generally respond to clinically recommended antimicrobial agents.

Haemophilus influenzae

H. influenzae and other Haemophilus species are examples of common clinical pathogens with special growth requirements. The widespread emergence of β-lactamase–producing strains has made rapid detection of β-lactamase production and, in turn, detection of ampicillin resistance in Haemophilis important objectives (102,115–117).

Antimicrobial resistance among clinical isolates of H. influenzae has been monitored in the United States, Canada, and Europe. In a comprehensive study (118) involving isolates in the United States, 20% of all H. influenzaeisolates were ampicillin-resistant by virtue of β-lactamase production. Enzyme-mediated resistance to ampicillin is approximately twice as common in serotype B strains (31.7%) as in non–B strains (15.6%). As a result of the production of the inactivating enzyme chloramphenicol acetyltransferase (119), occasional resistance has been noted with chloramphenicol, as well as resistance to tetracycline, trimethoprim-sulfamethoxazole, rifampin, and first-generation cephalosporins.

Because of the worldwide prevalence of β-lactamase–producing strains of H. influenzae, it is imperative that clinical laboratories routinely perform β-lactamase studies on all clinically significant isolates. Several media formulations have been employed for dilution or diffusion susceptibility testing with H. influenzae (65,120). Although individual methods may have been useful in individual laboratories, a common problem with these media formulations has been the complexity of their preparation and their opaque nature. Recently, a new simplified medium, HTM, has been developed; it avoids many of these problems (65). HTM is optically clear, stable, and reproducible from lot to lot and is currently available commercially from several manufacturers. The use of HTM has been advocated by CLSI (NCCLS) for both dilution and diffusion tests with H. influenzae (10). Guidelines for the interpretation of MIC and QC results with an H. influenzae reference strain are included in this chapter and are contained in Table 3.28.

In addition to H. influenzae, other problem organisms have been tested, as indicated in Table 3.28. Following is a discussion of current approaches to the testing of MRSA, N. gonorrhoeae, S. pneumoniae, S. viridans, L. monocytogenes, E. corrodens, and Chlamydia spp, along with appropriate references. In general, the use of a full incubation interval of 24 hours and the addition of 2% NaCl to cation-supplemented MHB have facilitated the testing of MRSA and have greatly enhanced the reliability of standard microdilution tests in detecting such organisms (121–123). Variable results have been obtained with different dilution methods for testing MRSA strains with cephalosporins, and many strains appear fully susceptible to cephalosporins, which is inconsistent with treatment results. This has led to the standard recommendation that cephalosporin tests not be reported for MRSA and methicillin-resistant S. epidermidis (11).

Neisseria gonorrhoeae

The changing antimicrobial susceptibility of N. gonorrhoeae is an example of how clinical laboratories have had to modify their approaches and strategies in susceptibility testing. N. gonorrhoeae has developed resistance to all of the agents that have been recommended for gonorrhea therapy. When penicillin was the recommended therapy for N. gonorrhoeae infection, most isolates were initially susceptible. From the mid-1940s to the 1970s, there was a 24-fold increase in the dosage of procaine penicillin (2 × 10⁵ units to 4.8 × 10⁶ units). Penicillin is no longer the recommended therapy.

The resistance of N. gonorrhoeae to antibacterials is due either to multiple chromosomal mutations or to R factor plasmids. In 1987, the increasing prevalence of strains with β-lactamase plasmids prompted discontinuation of the use of penicillin as a single-dose therapy. Determining resistance, which is primarily a laboratory responsibility, affects epidemiologic surveillance and patient care.

A standardized laboratory method for monitoring the susceptibilities of gonococcal isolates was formulated and has been recommended by CLSI (NCCLS) based on a multicenter laboratory study whose purpose was to standardize disk diffusion and agar susceptibility tests (124). The recommended test medium is GC agar base with a defined XV-like supplement. Three QC organisms are required: N. gonorrhoeae ATCC strain 49226 (CDC F-18), N. gonorrhoeae WHO strain V, and S. aureus ATCC strain 25923.

The publication of the multicenter guidelines does not alter the need or methodology for detecting penicillinase-producing N. gonorrhoeae. Penicillinase-producing N. gonorrhoeae strains may be identified by detection of β-lactamase with a nitrocefin substrate (Table 3.28). Strains of N. gonorrhoeae that have chromosomally mediated resistance to antimicrobial agents or plasmid-mediated resistance to penicillin and/or tetracycline may be detected by measuring their susceptibilities by disk diffusion (124) and are not discussed here. Agar dilution methodology is preferred to broth dilution, as N. gonorrhoeae is known to autolyze in liquid media.

Streptococcus pneumoniae

Throughout the world, there have been increasing reports of the relative resistance of S. pneumoniae to penicillin as well as to other drugs such as tetracycline, erythromycin, clindamycin, and chloramphenicol (16,17). Nonsusceptible strains have been isolated in the United States with increasing frequency; thus, routine testing of isolates of S. pneumoniae is probably warranted. Satisfactory broth dilution tests can be performed with S. pneumoniae, provided that the broth medium is appropriately supplemented by adding 5% defibrinated sheep blood to freshly thawed MHB in microdilution trays that have been stored for no more than 2 months at 20°C. Commercially prepared test systems are available but have limitations and need to be verified and validated (125). Tests for other streptococci are indicated in Table 3.28.

Special studies related to susceptibility testing of several other species have been performed and are noted here. MICs, MBCs, and killing curves have been studied for L. monocytogenes using TS broth (126,127). Pasteurella multocida was tested in microtiter panels using MHB supplemented with 10% horse serum (128). Vanhoof et al. (129) evaluated inhibition and killing of Campylobacter jejuni in MHB. E. corrodens susceptibility testing has been accomplished using MHB supplemented with 0.5% lysed sheep blood (130). Mycoplasma spp have also been studied (131) in macrotube systems and by microdilution susceptibility testing (132,133). Unlike conventional bacteria, which can grow in artificial media, Chlamydia trachomatis has been tested against a variety of antimicrobial agents by incorporating antibiotic dilutions in cycloheximide-treated McCoy cell cultures (134,135). Utilizing this approach, MICs and MBCs can be determined (see Chapter 7).

Nocardia spp and Actinomycetes

In large part, problems associated with the susceptibility testing of this group of microorganisms relates to the new directions in the methods used for their identification. In the last 15 years, newer approaches such as molecular testing using 16S ribosomal RNA sequencing and Matrix-Assisted Laser Desorption Ionization-Time of Flight (MALDI-TOF) methodologies have been adopted (136,137). Although broth microdilution is the recommended methodology, false resistance using this approach has been documented, specifically for Nocardia spp (138). The generally recommended antimicrobial agents for primary testing are amikacin, amoxicillin-clavulanate, ceftriaxone, ciprofloxacin, clarithromycin, imipenem, linezolid, minocycline, moxifloxacin, trimethoprim-sulfamethoxazole and tobramycin. Secondary testing compromises cefepime, cefotaxime, doxycycline, rifampin, and vancomycin. Susceptibility testing for the aerobic actinomycetes requires 2 to 5 days, contrasted to 24 to 48 hours for species of Rhodococcus.

Although there has been limited development of new antimicrobial agents and classes over the past 10 years, microorganisms have developed novel and, in some cases, multiresistant mechanisms to limit the clinical use of the available compounds. Several phenotypic and genotypic methods are available as screening tests, which do not result in an MIC but possess the necessary accuracy that confirmatory testing is not required and results can be used to guide antimicrobial therapy. Molecular, that is, genotypic, methods are reviewed by Hegstad et al. in Chapter 9 of this volume “Molecular Methods for Detection of Antibacterial Resistance Genes: Rationale and Applications”; phenotypic approach will be discussed here.

β-Lactamase Detection

Three iterations of this assay are available: acidometric, iodometric, and chromogenic; the latter being the most common. Results are typically available within 60 minutes. Although β-lactamase production inactivates ampicillin, amoxicillin, and penicillin for N. gonorrhoeae, it detects only one form of resistance and for chromosomally mediated resistance with PBPs, resistance can only be detected by disk diffusion or agar dilution methods. Because some β-lactamase–producing organisms (e.g., Staphylococcus) may produce detectable levels of the enzyme after induction, it is generally recommended that for serious infections requiring penicillin therapy an MIC be performed following the rapid phenotypic β-lactamase assay.

Detection of Methicillin Resistance: Methicillin-Resistant Staphylococcus aureus and Related Resistance Phenotypes

Rapid and early diagnosis of MRSA is necessary for prompt initiation of appropriate antimicrobial therapy. The most common method is culture modified more recently with the application of chromogenic agars (139). Other than molecular approaches there are no equivalent rapid “broth” phenotypic approaches. However, broth phenotypic resistance screening tests for S. aureus are extant and address mecA-mediated oxacillin resistance, inducible clindamycin resistance, and high-level murpirocin resistance (Table 3.29).

Rapid approaches for detection of methicillin resistance in S. aureus has become a major concern in hospital and community settings as the potential for this organism to become a multidrug-resistant pathogen capable of causing mild to severe infections poses a high risk of mortality (140). Although β-lactam antimicrobials are the drugs of choice for S. aureus whenever possible, glycopeptides—specifically, vancomycin—has become frontline therapy. The increased use of vancomycin associated with higher incidences of MRSA has resulted in reduced susceptibility to vancomycin.

In 1977, the initial strain of S. aureus with reduced susceptibility to vancomycin was reported in Japan (141). Strains considered to have reduced susceptibility are the following: vancomycin-resistant Staphylococcus aureus(VRSA) are characterized by MICs greater than or equal to 16 µg/mL; vancomycin-intermediate Staphylococcus aureus (VISA), characterized by MICs 4 to 8 µg/mL; and heterogeneously vancomycin-intermediate Staphylococcus aureus (h-VISA), defined as the presence of isolated populations of VISA with concentrations of organism at 1/10⁵ to 10⁶ susceptible to vancomycin. h-VISA appears to be the stage before the development of VISA. The presence of h-VISA significantly compromises the treatment of patients with bacteremia and frequently escapes detection in the clinical laboratory. Population analysis profile/area under curve (PAP-AUC) ratio is considered to be the gold standard for detection; however, due to its complexity, it is not feasible to routinely perform in clinical settings. An expedient alternative is the E-test macromethod (MET) (140).

Clindamycin Resistance in Streptococci and Staphylococci

In streptococci, clindamycin and/or erythromycin resistance may be associated with the erm genes, which enable production of macrolide ribosomal methylases or to the expression of the mef gene, which turn on an efflux pump targeting only macrolides. Alternatively, erm enzymes cause decreased binding of macrolides and lincosamides (e.g., clindamycin and streptogramin antibiotics—the macrolide-lincosamide-streptogramin [MLS] phenotype). The MLS resistance phenotype can be either induced or constitutively expressed. Inducible clindamycin resistance cannot be determined by routine susceptibility testing. A suggested approach is the use of a single-well microdilution test containing 4 µg/mL erythromycin and 0.5 µg/mL clindamycin. Any growth in the well is an indication of inducible clindamycin resistance; no growth indicates the absence of inducible clindamycin resistance (see Table 3.29).

Aminoglycoside (High-Level Resistance) in Enterococcus

Typically, enterococci are innately resistant to low levels of aminoglycosides owing to their facultative anaerobic metabolism, which limits the uptake of drug. Cases of serious enterococcal infections and endocarditis mandate the use of an aminoglycoside plus a cell wall–directed agent such as ampicillin, penicillin, or vancomycin, or specifically in the latter case, an aminoglycoside and streptomycin.

The antibiotic combination allows for the increased uptake of drugs and leads to the directed bactericidal activity against the organism in the absence of high-level aminoglycoside resistance (HLAR). To determine the potential for HLAR, CLSI recommends HLAR screening of enterococci with gentamicin and streptomycin using a broth microdilution method by determining the growth capability of the recovered enterococcal isolate in the presence of 1,000 µg/mL of streptomycin or 500 µg/mL of gentamicin in BHI broth; other approaches are also described (142). (See Tables 3.29 and 3.30.)

Vancomycin-Resistant Enterococci

Vancomycin-resistant enterococci, common colonizers of the gastrointestinal tract represent opportunistic pathogens in health care facilities. Two common patterns of enterococcal resistant are extant; both demonstrate elevated vancomycin MICs. The most clinically important type is vancomycin resistance associated with the acquisition of genomic elements in the form of a plasmid or other transmissible genetic element. Representation of this acquired trait is more frequently encountered in strains of Enterococcus faecium and E. faecalis harboring van A or van B genes that erode high levels of vancomycin resistance expressed as MICs greater than 128 µg/mL for the van A gene and lower MICs of 16 to 64 µg/mL for the van B gene. Another pattern of vancomycin resistance is intrinsic in nature and associated with van C and potentially other van genes; for example, van E, van G, and van L observed in species of Enterococcus gallinarum and Enterococcus casseliflavus. This expression of resistance results in low to intermediate levels typically in the range of 2 to 16 µg/mL. Guidelines for the susceptibility testing of enterococci to vancomycin are in the 2012 CLSI document (143).

Detection of Extended-Spectrum β-Lactamases in Enterobacteriaceae

Several genera among the Enterobacteriaceae and P. aeruginosa are capable of producing β-lactamases referred to as extended-spectrum β-lactamases (ESBLs), which have the capacity of hydrolyzing penicillins, aztreonam, and the cephalosporin groups encompassing the extended spectrum moieties as cefotaxime, ceftriaxone, ceftizoxime, and ceftazimine. Guidelines are available that specify screening and confirmatory approaches for detecting ESBL production in E. coli, Klebsiella pneumoniae, and Proteus mirabilis (144). (See Table 3.31.) ESBL screening should not be performed for Enterobacter spp and Serratia spp, which produce Amp C–type hydrolyzing enzymes; ESBL screening is not advised because false-negative results can occur. The need to screen and confirm the presence of ESBLs was necessitated because standard MIC (and disk diffusion) susceptibility testing were not uniformly capable of identifying isolates capable of producing ESBLs. As ESBLs are typically inhibited by clavulanic acid, this property was used to detect ESBLs and interpret that presence as potential for conferring resistance to all penicillins, cephalosporins, and the monobactam aztreonam. The establishment of new lower interpretive criteria for the three aforementioned drug groups based mainly on PK and pharmacodynamic (PD) data permitted clinical laboratories to abandon screening tests and that testing results for the three drug groups be reported as tested. The European Committee on Antimicrobial Susceptibility Testing (EUCAST) has adopted similar breakpoints for the cephalosporins (see Tables 3.32 and 3.33) and recommends that laboratories continue to screen and confirm ESBL production due to limited clinical data and that cephalosporin reporting results be diminished by one category of interpretation (i.e., “susceptible” to “intermediate”) if ESBLs are detected (145).

Detection of Carbapenemase Activity in Enterobacteriaceae

Enzymes that hydrolyze the carbapenem class of antimicrobials, doripenem, ertapanem, meropenem, and imipenem—referred to as carbapenemases—usually hydrolyze other currently used β-lactams with the exception of aztreonam. Carbapenemases have been identified in a wide range of gram-negative genera and are subdivided into three classes composed of serine class A (including KPC, SME, IMI, GES, and NMC); class B enzymes, the metallo-β-lactamases (VIM, IMP, and NDM); and class D OXA enzymes (146). KPC is recognized as the most frequently identified class A carbapenemase in the United States and can be found in the Enterobacteriaceae and in P. aeruginosa (147). Metallo-β-lactamases are typically detected in Acinetobacter spp and P. aeruginosa; however, the NDM group is widespread among Enterobacteriaceae, particularly K. pneumoniae. The OXA group is associated with Acinetobacter spp but has also been reported in the Enterobacteriaceae.

Other than in 2009, the modified Hodge test was the recommended methodology for the detection of carbapenemase activity in Enterobacteriaceae (148). The assay is an agar diffusion methodology, which uses the reduced activity of a carbapenem, ertapenem, or meropenem observed by an indentation of the zone of inhibition against known stock strain of carbapenemase-producing K. pneumoniae. This test was originally recommended when carbapenem MICs were elevated or resistant. In 2010, the CLSI lowered the breakpoints for carbapenems to identify carbapenemase-producing strains, which would test as intermediate or resistant to this antimicrobial group. Use of the revised breakpoints eliminates the need to routinely perform the modified Hodge test, although such testing may be of value for epidemiologic or infection control.

An overview of the guidance for detection of ESBLs and carbapenemases as suggested by Livermore et al. (149) appears in Table 3.34.

Detection of Plasmid-Mediated AMPC-Type β-Lactamases

This group of β-lactamases, in some cases chromosomally mediated, are produced by a wide range of gram-negative species. Although detection of this group of enzymes is deemed important for epidemiologic and infection control initiatives, no sufficiently standardized testing regimens has been developed to use a screening approach.

Direct Microscopy and Observation of Bacterial Morphology

Early in the use of penicillin, it was noted that cultures of either spinal fluid or blood taken from individuals who had been treated with penicillin demonstrated aberrant forms in these clinical specimens. Observations of morphologic alterations in bacteria produced during early exposure to various antimicrobial agents were recorded by several investigators (150,151). Growth can be observed by a variety of means, including light, phase-contrast, and electron microscopy, with different changes being produced by an antibiotic, depending on the bacterial species and the mode of action and concentration of the agent. Antibiotics, especially those of the β-lactam class, which interfere with cell wall synthesis and bind to the lytic PBPs (see “Minimal Bactericidal Concentration”), can produce various filamentous forms or protoplasmic enlargements (152). Similar morphologic alterations can be seen in inocula exposed to concentrations that are below the MIC but that nonetheless modify the characteristics of the affected bacterial population (153).

The minimum amount of antimicrobial agent necessary to induce such alterations has been termed the minimal antibacterial concentration (MAC) (153,154). When visualized by light microscopy, this approach represents a rapid method because the changes may occur rapidly, occasionally within minutes. However, these observations are qualitative, the methods are not standardized, and the observations are labor-intensive.

Clinical data demonstrating the relevance of a given alternative antibiotic susceptibility test method in the care and management of infectious disease are based primarily on the MIC of the agent against the infecting organism. Additional investigation is needed to establish the interpretive significance of a standard derived from the new approach. Thus, in the MAC approach, one would have to study the MAC/MIC ratios (153,155) to determine the MAC’s clinical utility.

Early Reading of Conventional Tests

Like in the early reading of disk diffusion tests, where the kinetics of growth and zone formation are fairly well stabilized for many drug–bug interactions within 8 hours of incubation and frequently within 5 hours (156) or 6 hours (88), results of rapid or early reading can be estimated in liquid media. It is suggested, however, that early estimated results be confirmed by a later reading or by retesting with conventional incubation intervals. Lampe et al. (157) determined the accuracy of early reading compared with standard ICS macrotube dilution (i.e., 3 hours in comparison with the conventional 18-hour incubation time). Values from the conventional incubation were at least four times higher than those from the early 3-hour reading for 43% of the strains tested. Agreement was improved with the 8-hour incubation period. When procedural modifications were initiated for the purpose of producing rapid results that would conform to conventional end points, comparisons of different media, shaking versus static cultures, and reading by particle count versus visually did not eliminate the discrepancies (157). However, increasing the inoculum size to possibly 10⁷ CFU/mL produced better agreement, with 14% of the test isolates demonstrating fourfold dilution errors (Table 3.35). Because this approach is not applicable to all species, especially in view of variable mechanisms of resistance, additional clinical investigations of these interpretations are warranted. Until such investigations for establishing clinical guidance in earlier time frames are completed, the focus will be on the reference overnight method.

There have been attempts over the years to develop methods that could enhance the apparent growth of bacteria and thereby reduce the conventional overnight incubation required to determine an MIC. Investigators have used colorimetric indicators and leuco dyes to amplify the macroscopic observation of turbidity. In macrodilution assays, phenol red was added to assay tubes containing glucose and yeast extract to enhance growth. If the test organism grew utilizing the glucose, tubes would turn yellow as a result of the pH change. If the antimicrobial agent inhibited the organism, the indicator would remain red. Nonfermenter organisms could be detected by an orange-red color if they grew in the antibiotic-free tubes.

Redox indicators such as resazurin, triphenyltetrazolium chloride, and methylene blue have been used as indicators of growth (158). Because these dyes may be antibacterial, the strategy was to add them to the control tube after a brief (3- to 5-hour) incubation period. If sufficient growth was then detectable, the indicator would be added to all assay tubes. Bartlett and Mazens (159) enhanced the sensitivity of triphenyltetrazolium chloride by the addition of phenazine methosulfate, which accelerated formation of the red formazan precipitate resulting from growth of the bacteria. The indophenol reagent dichlorophenolindophenol is another redox dye that has been tested (93). A proprietary tetrazolium-type reagent (Alamar blue) incorporated into commercially prepared antibiotic-containing testing panels has received some attention (160).

In a unique approach, bacterial respiration was detected utilizing the respiration of hemoglobin as an indicator. A 3% suspension of red blood cells (outdated human or animal) served as the indicator (161).

The rapid approaches described in these early reports have not been accepted by and are not applied in clinical or reference laboratories. Had they received some commercial impetus and if ample reference and QC measures had been developed to ensure reliability, they might have achieved acceptance.

Alternative Rapid Instrumental Methods

Detection of Bacteria and Bacterial Products

In the search for rapid methods for determining the interaction of antimicrobial agents and organisms, microbiologists have examined the products of bacterial metabolism (intermediate and end products) as well as the interaction of the organism with various energy sources. In addition to the more complex products of bacterial metabolism detectable by gas chromatographic methods (162), there are the more immediate end products of glucose degradation (i.e., CO₂, water, energy, adenosine triphosphate [ATP], and heat). Thus, if glucose is tagged with radioactive carbon or some other substrate is similarly tagged, then the CO₂ released via bacterial metabolism is radioactive and can be detected by radiometry (163–166). Alternatively, the CO₂ produced can be monitored by infrared spectroscopy to detect growth (266). Also, the ATP produced by metabolism can be measured by the luciferin-luciferase reaction, and the energy quanta derived from this interaction can be measured with an ATP luminometer (167). Similarly, heat in the form of nonutilizable energy derived from the enzymatic degradation of glucose can be detected (Fig. 3.8) and measured by microcalorimetry (27). As bacteria replicate in the growth medium, they decrease the measured electrical resistance (impedance in alternating-current circuits) in the medium (Fig. 3.9). Impedance has been measured in bacterial cultures by passing a high-voltage, alternating-current signal through the medium and monitoring its effect (168–170). All of these approaches are bacteriologically sound, in that any method used to detect bacterial growth can be applied to the measurement of the interaction of antimicrobial and bacterium and thus to the testing of susceptibility (27,171).

Of the several methods noted here, two persist in commercial instrumentation suitable for clinical laboratories: radiometry and ATP luminometry. Only the former is currently applied to antimicrobial susceptibility testing, specifically with mycobacteria (172,173,269).

Another approach to the detection of bacteria and, in turn, the determination of susceptibility involves measuring bacteria interacting with some component of the electromagnetic spectrum. Theoretically, one can measure single particles (individual bacteria) or, for better accuracy, a population of particles (bacteria). The origin of using densitometry (turbidity and visual turbidity) (Fig. 3.10) can be traced to the hallmark works of McFarland (174) and Longsworth (175). Specialized light-scattering methods (Fig. 3.11) using coherent light emanating from a laser have also been applied (176,177). Infrared light (126) and ultraviolet light have been used as energy sources to detect bacteria. Radiation of the latter type interacts with nucleic acids at 260 nm. The uptake of thymidine, a DNA precursor, by microorganisms in the presence and absence of antimicrobial agents has been assessed by Amaral et al. (178). In this assay, labeled thymidine (with either tritium or carbon as the label) is used to determine susceptibility profiles directly from a patient’s specimen, thereby obviating the need for a primary culture and the time delay associated with the initial culture. The assay determines the amount of radioactive thymidine incorporated into DNA compared with control and antibiotic-containing cultures. Clear drawbacks of this approach include the potential contamination of the workplace and technical staff and the costs associated with the necessity to routinely monitor the environment and personnel and dispose of materials.

Flow Cytometry

A recent significant development in antimicrobial susceptibility testing is the application of flow cytometry (179). The technology of flow cytometry encompasses aspects of immunology, biochemistry, molecular biology, and histology. Any cellular analyte or metabolic event that can be tagged with a fluorescent dye is a potential candidate for analysis by flow cytometry. Flow cytometers are complex instruments linked to powerful computers still largely dependent on manual procedures. The essence of current multiparameter flow cytometry resides in the capability of the instrument to analyze individual cells (and cell types) of a subpopulation within a larger heterogeneous population without the need for separating or isolating the subpopulation. During flow cytometric analyses, cells in suspension are pushed (flow) single file through a flow cell (Figs. 3.12 and 3.13), where the cells are exposed to monochromatic light emanating from a laser beam, usually of the argon-ion type (180,181). Two cellular parameters derived from the scatter of the laser light are measured: size (forward light scatter) and internal complexity or surface irregularity (right-angle scatter). Detectors in the flow cytometers are sensitive to the colors emitted by fluorochrome-tagged antibodies or fluorescent tracers after excitation by the laser beam.

The technique, as applied to antimicrobial susceptibility testing of microbes, exposes the organism to an antimicrobial agent for 2 to 6 hours at an appropriate incubation temperature. After exposure, organisms are stained with a fluorescent stain like ethidium bromide, which results in increased fluorescence due to breakdown of the cells. The increased fluorescence is then interpreted as death and could be used to determine the MIC. This approach has been applied to M. tuberculosis (182). There is no commercial instrument exclusively for this specific application. The current drawbacks of using flow cytometry for bacterial susceptibility testing are the cost of the system and the limited data accumulated for bacteria.

Automated Antimicrobial Susceptibility Test Systems

During the 1980s, clinical microbiology laboratories increased their use of commercial instrument-associated antimicrobial susceptibility testing methods. The growing acceptance of these commercial instruments by laboratories parallels progress in electronics, robotics, and microcomputers—progress that permitted manufacturers to develop instrumentation able to identify routine gram-negative and gram-positive bacteria of medical importance, determine the associated antimicrobial susceptibility, and combine the results into a single report. The broth microdilution susceptibility tests and associated identification component have become the most popular of the systems currently available to clinical microbiology laboratories in the United States (183).

Background, Development, and Description of Automated Susceptibility Test Systems

The earliest automated and probably rapid susceptibility test system was the TAAS (Technicon Automated Antimicrobial Susceptibility) system. It was developed in the early 1970s by Technicon Instrument Corporation (Tarrytown, NY). It was never marketed. In this system, an antimicrobial agent was delivered by dropping antibiotic-impregnated disks into broth. Through the process of elution, the antimicrobial agent was free to act in a short period of time. After bacterial inoculation and growth for 3 hours in the presence of the single antimicrobial concentration, growth was compared with a control broth without antibiotic. A growth index was calculated based on the ratio of the growth in the presence and absence of the antimicrobial agent. From this index, an evaluation was made as to whether the bacterial population was susceptible (or resistant) to the antimicrobial agent. Instruments that immediately followed, such as the Autobac system (developed and marketed by Pfizer Diagnostics), also used antibiotic elution from paper disks as a means to initiate antimicrobial activity.

Subsequent systems utilized the microtiter 12 × 8 (96-well) format, in which antibiotic was prepared and frozen until used. In some cases, a lyophilized dry powder was used. Clearly, there are advantages associated with dry powders, such as increased shelf life and storage flexibility. Systems utilizing microtiter formats became increasingly automated, to such an extent that the inoculation, incubation, and readout could be handled without human intervention. Once microcomputers with expanded memory became available, along with the capability of printing susceptibility results with bacterial identification and associated patient demographics, data management systems that could be used for the generation of cumulative susceptibility profiles were developed.

An abbreviated list of automated susceptibility systems available during the past two decades is presented in Table 3.25. As noted, most systems are no longer manufactured, and the commercial sponsors of the viable systems have changed.

Of the several antimicrobial susceptibility testing systems listed in Table 3.25, all but one use visible light as a measuring parameter (184). The Sensititre system uses fluorometric monitoring for determining the interaction of antimicrobial agent and bacterium (185). The systems that detect early growth with limited incubation (less than 6 hours) are the Autobac, the Avantage, and the AMS Vitek and Sensititre systems. If one considers the operational definition of rapid as 4 to 6 hours, then only the Autobac and Avantage systems fall into this category. All of the other systems require 6-hour or overnight incubation. However, the effects of limited incubation intervals, especially with β-lactam drugs, warrant serious evaluation (186). The accuracy of these systems (plus or minus one dilution) ranges from 87.6% to 98.3% (124).

It is not productive to describe in detail the varied commercial systems that were available, because frequent changes and corporate direction have modified the design and features of these instruments. Table 3.36 outlines contemporary instruments currently available and widely used. The descriptions of the two systems that follow—the MicroScan system, originally developed by Baxter, and the system from Vitek/McDonnell Douglas, now Vitek bioMérieux (Hazelwood, MO)—represent different approaches to automated susceptibility testing.

The MicroScan system developed in the 1980s uses the conventional 96-well format or an extended design coupled with fluorogenic or standard substrates to detect bacterial growth. The system includes a large computer-controlled microprocessor that incubates standard microdilution trays and interprets biochemical and/or susceptibility results with either a fluorometer for fluorogenic substrates or a conventional photometer for standard plates. (Panels with fluorogenic substrates have been discontinued.) The WalkAway model is available in two sizes, with a 40- or 96-test panel capacity. It consists of a large, self-contained incubator/reader with humidifier, a carousel that rotates towers containing the panels, a bar code scanner, photometers (spectrophotometer/fluorometer), and associated robotic mechanisms to move panels and perform designated computer-controlled robotic steps to access and position trays, add reagents, and position trays for reading. The instrument is associated with a microcomputer, a video display terminal, and a printer. The database management system is capable of producing a patient report and storing that information to subsequently reproduce cumulative reports, epidemiologic reports, and antibiograms. Although the instrument discussed is automated, preparation of the inoculum is done individually and manually using a seed trough, which is transferred to the test panel with a rehydrator/inoculator (RENOK). This approach is more conventional, using the microtiter format with multiple antibiotic concentrations directed by microcomputer-controlled robotics (except for inoculum preparation) to achieve a final MIC result.

In contrast to the microtiter panel approach platform is the specialized design and concept of the Vitek system. The essence of this system is the reagent card, a small thin plastic unit that contains 30 wells or microcuvettes in the Legacy version and 45 wells in the Vitek II version. The wells are connected by capillaries in which bacterial test suspension passes for rehydration and inoculation. The cards are available with a variety of predetermined configurations of antimicrobial agents and reagents to identify gram-positive or gram-negative organisms. The MIC end point is determined by an algorithm from results of testing one to five antimicrobial concentrations. Cards are stored at 4°C, with a shelf life of approximately 12 months. The system includes integrated hardware consisting of a filler/sealer that serves as the inoculator and sealer for up to 10 cards; a reader/incubator that contains the robotic system to move the cards on a timely basis from a carousel to a position in the instrument where optical density or biochemical reactions are determined by a photometer; and a computer module that contains software, a video display terminal, and a printer. Systems are available in different capacities and can hold from 30 to 240 cards. Growth is determined turbidometrically at hourly intervals for up to 15 hours and is compared to the baseline and expressed as a ratio. Normalized linear regression of the growth is used to calculate a best fit to determine the MIC. The user can opt to print susceptibility results as discrete MICs or qualitative breakpoint results (susceptible to resistant).

Each of the systems described here can be linked to the main laboratory information system through a standard RS232 interface. The data management system is designed for storing and retrieving data for laboratory pharmacy and infection control purposes as well as printing chartable patient reports.

In addition to the instruments listed in Table 3.36, there are available several semiautomated devices that only turbidometrically or fluorometrically read/scan test panels inoculated manually and incubated overnight (18 to 24 hours). These include the AutoScaptor (Becton, Dickinson and Company, Franklin Lakes, NJ), MicroScan autoSCAN-4 (Dade Behring Inc, Deerfield, IL), mini API (bioMérieux, France), and Sensititre AutoReader (Trek Diagnostics Inc, Cleveland, OH). Also not included in Table 3.36 are the several computer-assisted semiautomated devices that measure disk diffusion zone diameters by image analysis (see Chapter 2). This latter group of instruments interprets zone diameter readings according to a user-selected database for reporting S, I, and R categories. One system, BIOMIC (Giles Scientific, Santa Barbara, CA), determines MICs by regression analysis of measured zone diameters. Clear advantages of this instrument group are that they generally provide results that are more reproducible than those achieved by manual observation, measurement, and recording and that they can be interfaced to laboratory information systems, thereby reducing the potential for transcription errors in patient reports.

Advantages of Automated Systems

A perceived advantage of the automated approach to antimicrobial susceptibility testing is the apparently more efficient use of robotics, rather than humans, in executing these tests. However, the gain in labor savings is not extensive. Perhaps a more meaningful advantage is the reproducibility of results obtained by using this type of instrumentation. Because procedures for inoculum preparation, the specified duration of incubation, and the assessment of growth are standardized in these systems, subjective components are removed and intralaboratory and interlaboratory standardization is readily achieved. For high-volume laboratories that demand high throughput, the time required for reading and interpreting routine susceptibility tests is diminished. The associated capability of performing identifications is an additional advantage. Considerable labor savings are also derived from the potential to establish a link between the computer and the laboratory information system. This capability precludes technologist error in transcribing, sorting, and entering results in individual reports.

Limitations and Problems of Automated Systems

Systems that mimic and merely robotize a multiwell approach to performing conventional MICs seem to have only minor disadvantages. However, when the systems associated with these instruments offer limited testing panel capacity are not applicable to all groups of bacteria and fail to incorporate QC end points that are on scale (i.e., within the MIC range of the test panel), their flexibility and precision are questionable. Furthermore, the capital outlay required for purchase and the high cost per test for reagent rental acquisition need to be assessed and compared with the expense of manual test methods. If the clinical impact significantly improves the quality of patient care and reduces hospital costs overall, then the financial costs of the testing itself become irrelevant.

The design of automated instruments that utilize extremely small volumes, modify the inoculum, and reduce the incubation interval has created several problems in the detection of resistance. The detection of type 1 inducible β-lactamase resistance associated with Citrobacter, Enterobacter, Serratia, Providencia, Pseudomonas, and indole-positive Proteus strains is hampered by short detection (incubation) intervals. Owing to their inherent nature, inducible-type resistors are missed because a longer incubation is required for induction and expression and subsequent detection of spontaneous mutants.

Because of the failure to detect emerging vancomycin resistance in enterococci, the FDA had imposed prohibitions for automated testing of enterococci with vancomycin. Similarly, these restrictions have been applied to S. pneumoniae susceptibility testing. As noted earlier, the limitations associated with short incubation times contribute to these problems.

Other specific problems related to automated susceptibility systems have been encountered. These include false resistance to aztreonam, especially for Proteus and Morganella spp, and false resistance to imipenem associated with Pseudomonas spp.

A serious limitation of contemporary instrumentation concerns the newly recognized need to verify methicillin resistance. This is especially pertinent when testing coagulase-negative staphylococci. Verification can be accomplished by molecular detecting of mecA or its surrogate encoded protein product, PBP-2A, by latex agglutination. MecA detection is the gold standard. Alternatively, CLSI (NCCLS) has previously recommended a cefoxitin agar diffusion methodology (11).

During the development and evaluation of each commercial automated systems, reports have cited problems. As problems are identified, the manufacturer has attempted to address then by modifying the system by adjusting the growth support and cation content of the growth medium, modifying the reagents, and revising the software and/or algorithms associated with the optical reading device. Due to ongoing modifications, it is difficult to compare and evaluate the accuracy of the systems. As clinical microbiology laboratories come to understand the limitations of these systems, they may rethink the apparent benefit of automation and return to classical approaches, which are more flexible, less fraught with resistance problems, and less costly.

Verification of Antimicrobial Susceptibility Test Systems

When the laboratory considers endorsing a newly acquired automated test system or transitions from an extant system or conventional microdilution method to a new system, it is critical that the test performance of the new system is verified. Verification is accomplished by using the new or revised test method in parallel with a reference method with a known and satisfactory level of performance. The evaluation of susceptibility test methods should be done using a distribution of organisms typically encountered at the institution, ideally including susceptible and resistant strains for each antimicrobial agent. Guidelines are available, but generally the evaluation should be designed to include at least 100 strains and detect three types of errors categorized as very major (reference method R; new system S), major(reference method S, new system R), and minor (reference method R or S, new system I). Very major errors should be less than or equal to 3% and major and minor errors should less than or equal to 7% (187).

Clinical Impact of Automated or Rapid Testing

Early evaluations of automated systems now more than 20 years old have failed to show that rapid susceptibility tests produce a sustained significant positive impact on patient care. In studies of patients with bacteremia, Doern et al. (188) found that, among 173 patients who were receiving antibiotics, rapid susceptibility testing indicated a change in therapy for 48 of these patients. For 32 of the 48 patients, the change in therapy was made 24 hours earlier as the result of 1-day earlier testing. Trenholme et al. (189) found that, for 226 bacteremic patients, rapid automated susceptibility results were associated with a greater likelihood of administration of appropriate antimicrobial therapy, a change to more effective therapy, and/or use of less costly therapy. These studies failed to find a change in patient outcome.

In a follow-up study, Doern et al. (190) evaluated two controlled patient groups in a tertiary care institution. One group was tested with conventional overnight microbiologic procedures for identification and susceptibility testing. The other group had the benefit of rapid, same-day procedures. The investigators found that, with regard to mortality rates, the patients receiving the advantage of rapid, same-day testing experienced a lower mortality rate (8.8% vs. 15.3% in the conventional testing group). Other advantages associated with rapid testing were fewer laboratory studies, fewer days of incubation, fewer days in critical care units, and shortened times prior to modifications in antimicrobial therapy. In a more recent study, Barenfanger et al. (191) documented clinical and financial benefit (reduced length of stay) of rapid identification and antimicrobial susceptibility testing.

Schifman et al. (192) determined that a significant issue impacting patient outcome was the slow reporting of and response to results in the postanalytical phase of testing. These authors found that the value of susceptibility test results in initiating or modifying therapy was reduced by inefficient reporting practices. In many institutions, getting clinicians to respond promptly to actionable health care information is an ongoing quality improvement concern.

Rapid methods to determine susceptibility (actually resistance) have been used and no doubt will continue to be used as an adjunct to conventional susceptibility testing. Mention has been made of the use of β-lactamase testing using nitrocefin disks and the detection of chloramphenicol resistance by chloramphenicol acetyltransferase assay (119,193). Tests modeled on this strategy have been designed using the breakthroughs in recombinant nucleic acid technology to promptly assess resistant mechanisms (194) (see Chapter 9). For example, by testing for the presence of a plasmid product (an enzyme) or the nucleotide sequence that directs the synthesis of the product, a laboratory can report results within hours. Several nucleic acid probes for detecting and identifying specific infectious agents are already FDA-approved, and they are useful in the clinical laboratory. At recent count, 106 plasmids have been identified, and the gene products probed can be used to detect resistance for 16 classes of antimicrobial agents (195,196). Recently, York et al. (197) evaluated the molecular detection of mecA by polymerase chain reaction and the standard method for determining methicillin resistance in coagulase-negative staphylococci. Since mecA encodes for a low-affinity PBP, PBP-2a, one could alternatively detect this protein product using a commercially available latex agglutination kit. It can be anticipated that commercial probes for resistance markers may be available in the near future. This approach to susceptibility testing (actually determining resistance) can be extremely useful, especially in therapeutic situations in which meningitis is present. A note of caution: the presence of a particular nucleotide sequence that indicates the potential for producing the plasmid-mediated resistance product does not always mean that the product is expressed by the bacterium.

Indirect guides to determine the susceptibility of S. viridans group streptococci associated with endocarditis have been proposed (198). Studies have shown that the quantity of glycocalyx produced by these organisms is correlated with the size of infected cardiac vegetations and resistance to antimicrobial therapy (276). Other approaches to screening resistance of selected pathogens using antimicrobial agents incorporated into agar have been used successfully.

It is clear that in contemporary medical practice, antimicrobial susceptibility testing is necessary component for the delivery of effective, cost-efficient medical care. The MIC is of major import in the comparative evaluation of antimicrobial agents and in the ability to predict their clinical effectiveness by gauging bacteriologic outcome. Ever since Fleming’s discovery of the activity of penicillin and the subsequent development of antimicrobial agents, methods have been evaluated to measure and define the true value and meaning of this interaction—the MIC. The testing of the interaction between microorganism and antimicrobial agent is clinically applicable when it defines a breakpoint that parts the susceptible from the resistant. Test conditions and parameters can indeed modulate the outcome of test results, as discussed. In basic microbiology research laboratories, the end point of a standardized test, with the resultant MIC, yields an operational definition that identifies the organism tested as susceptible to the lowest concentration in the series. This does not require further clarification. However, in clinical microbiology laboratories, the medical relevance of that susceptibility test result is paramount. The variables and considerations that form the basis of the recommendations for the development of susceptibility testing interpretive guidelines were discussed in a landmark report by Ericsson and Sherris (6) and formed the basis for the interpretive categories recommended by Bauer et al. (7) and the CLSI (NCCLS) documents on dilution susceptibility testing (10). A commonly held misconception is that the MIC and its interpretation are based solely on the ratio of the MIC and the peak achievable serum level. This is incorrect. In fact, the interpretive guidelines, as evaluated by CLSI and EUCAST, although somewhat different are based on a tripartite set of databases, as originally recommended by Ericsson and Sherris (6), Bauer et al. (7), and CLSI (NCCLS). The database for evaluating interpretive standards is dependent on the susceptibility of isolates in a large population distribution, the clinical pharmacology of the drug, and the drug’s clinical efficacy. When examining the MIC of new drugs for isolates or reviewing the efficacy of older drugs, the results are compared with the mode (or modes), range, and character (i.e., unimodal, bimodal, or skewed) of the distribution of MICs for populations of strains of the same species being tested. These ranges are then compared with the distribution of MICs of other species within the same group for the same drug and other agents of the same class.

In evaluating the clinical pharmacology of the drug, one includes the range, peak, mean, and trough serum levels expected from the variety of dosage schedules and also considers other PK parameters such as protein binding, volume of distribution, tissue level, and level of the active drug in urine. The range of mean serum levels rather than peak values more appropriately reflects overall tissue levels and avoids inappropriately amplified therapeutic ratios that result from comparing transient, maximally achievable peak serum levels with the MIC of the organism. Lastly, data are collected and evaluated from prospective clinical investigations that reflect the in vivo response to treatment of patients with specific infections caused by various strains of species with known MICs. Guidelines for performing these types of experiments have been included in a CLSI (NCCLS) document (199).

The problems of developing the interpretive breakpoints are usually more difficult for disk diffusion assays, where one has to establish the appropriate disk mass to produce zones of inhibition of a range usable in clinical microbiology laboratories. This is often done initially with customized or handmade disks and then confirmed later with commercially prepared formulations. The usual practice is to construct regression lines of a zone of inhibition versus MIC for about 500 separate isolates covering the spectrum of the new antimicrobial agent. These studies often include one or more comparative agents of the same class as the test agent. The bacterial strains used are obtained from geographic areas within the United States and represent the major routine isolates that would be expected to be treated with the agent under development. Breakpoints are selected based on the pharmacokinetics of the agent in humans, the ability to separate resistant and susceptible bacterial species, and the minimization of major errors of interpretation (false-susceptible or false-resistant). Additionally, for disk development, the paper containing the antimicrobial reagent has to be evaluated for content and stability over time and under storage conditions.

Differences in breakpoint determinations are extant between CLSI and EUCAST. These differences can readily be seen in Tables 3.32 and 3.33. Table 3.32 documents comparisons for 14 drug classes against five gram-negative microbial groups. Generally, EUCAST data demonstrate lower MICs than CLSI with the exception of the carbapenems. Of note is the threefold lower MIC listed for cefepime (≥4.0 µg/mL vs. ≥32.0 µg/mL) than CLSI. These variations are highlighted as a result of the 2011/2012 changes by CLSI as related to the detection of ESBLs.

For gram-positive bacteria (Table 3.33), notably Staphylococcus, there is dichotomy among the standard organizations for vancomycin, EUCAST indicating a threefold lower breakpoint for S. aureus and a fourfold lower concentration for the coagulase-negative staphylococci.

Over time, the criteria that have been used to establish breakpoints have changed, and various countries and organizations have adapted differing criteria (see Chapter 1). Early in antimicrobial susceptibility studies, it was observed that all strains of a bacterial species do not demonstrate the same degree of susceptibility and that the susceptibilities (i.e., MICs) for a population of strains are distributed according to a bimodal or normal distribution. Eventually, national committees and organizations such as CLSI came to conclude that in the derivation of a breakpoint, the PK properties (e.g., drug distribution in the host) and PD properties (e.g., activity against the infecting agent) of the antimicrobial compound indicated that the MIC (interpreted as susceptible) would suggest the possibility—perhaps the probability (not quantitatively defined)—of antimicrobial therapy success and resistant would connote failure. It is of interest to note the differing approaches used in various countries by the empowered organized committees that oversee or regulate antimicrobial susceptibility testing. The Dutch committee for regulating antimicrobial susceptibility testing, the CRG, states, “Micro-organisms are categorized susceptible (S), as opposed to resistant (R) to an antimicrobial agent; if concentrations of the non-protein bound fraction in vivo, based on the dosing regimen proposed, are above the MIC of a microorganism for a sufficient time-period in order to eradicate the microorganism, and thereby cure of the patient can be reasonably expected.” In contrast, the British Society for Antimicrobial Chemotherapy uses this formula: breakpoint concentration = C_max · csf (e.t), where C_max is the maximum serum concentration following a stated dose at steady state, e is the factor by which the C_max should exceed the MIC, f is a protein-binding factor, sis a reproducibility factor, and t represents the serum half-life. Clearly, organized groups have used varied data and criteria in establishing breakpoints, although the trend is toward the inclusion of PK and PD information.

In the United States, the FDA and the CLSI Subcommittee on Antimicrobial Susceptibility Testing have the responsibility for the development of standards that promote accurate and reproducible antimicrobial susceptibility testing and appropriate reporting. Data reviewed for establishing susceptibility breakpoints include in vitro drug characteristics, necessary distributions of microorganisms, PK/PD parameters, and the correlation of test results with outcome statistics.

As suggested, organizing committees in several countries are now using PK/PD relationships in their breakpoint estimations. For many drugs, data were accumulated documenting the relationship between drug concentration and effect. Three major characteristics of the concentration time curve (AUC) of a drug have been delineated: concentration (C_max), area under the AUC, and the time the concentration remains above the MIC (T>MIC) (Fig. 3.14). From these are derived PK/PD indices by relating the drug parameter to the MIC: AUC/MIC, C_max/MIC, and T>MIC. Studies of animal models of infection show an evident relationship between the PK/PD indices and efficacy. The setting of these breakpoints is derived from highest generally achieved indices: 100 to 125 for the AUC/MIC and 8 to 12 for the C_max/MIC (19). Examples of PK/PD relationships and breakpoints are shown in Table 3.37.

Although it was generally considered instructive and meaningful to classify antimicrobial agents as bacteriostatic or bactericidal (all antibiotics are bacteriostatic but not necessarily bactericidal), contemporary interpretation identifies two patterns of bacterial killing: time-dependent killing or concentration-dependent killing. Time-dependent killing agents are characterized by the PD parameter T>MIC, whereas concentration-dependent killing agents are characterized by C_max/MIC or AUC/MIC. Several antimicrobial classes classified by these PD parameters are listed in Table 3.38.

Although the MIC of a particular strain of microorganism recovered from a patient can be considered constant and reproducible (it may differ from other strains of the same species), it is known that the PK parameters of absorption rate, volume of distribution, and clearance can differ among individuals. Drusano and colleagues (202) acknowledged these variations and presented an integrated approach to population pharmacokinetics and microbiologic susceptibility information. Statistical simulation Monte Carlo (using a computer-generated program that integrates variously attainable PK/PD indices and MICs) can provide insight into the proportion of the population who can achieve an effective AUC/MIC and for whom clinical success can be quantitative by predicted (200–202). Monte Carlo simulation incorporates a probability function to generate random AUC and MIC values from a sampling distribution. Thousands of single-point estimates are made and their probabilities plotted (Fig. 3.15). Whereas single-point AUC/MIC estimates provide information on what is possible, Monte Carlo simulations can determine probability. Thus, one could predict the probability of achieving a targeted AUC/MIC ratio at different MIC breakpoints (Fig. 3.16).

Monte Carlo simulations and other statistical approaches, such as analysis using Classification and Regression Tree (CART), a statistical computer program for predicting the probability of a clinical response to infection by a microorganism as a function of the peak-C_max/MIC ratio (203), are and will be useful in developing more meaningful clinically applicable breakpoints. They do not enhance the value of the clinical laboratory providing comparative MICs for an infective organism with only an S, I, or R interpretation. What would prove applicable in light of this discussion is to provide the clinician with the number of log₂ concentrations below the S breakpoint concentration that was achieved; indirectly, this would be relative to AUC/MIC and C_max/MIC.

These considerations have led, at least in the United States, to the separation of data into three (or possibly four) interpretive categories: susceptible, intermediate (or “moderately susceptible”), and resistant. Susceptible organisms attain MICs that fall into the susceptible range of MICs and would generally be below serum or tissue levels of the drug achievable with usual dosage regimens; the implication is that infections attributable to such organisms may be appropriately treated with that drug. Strains classified as resistant would not be inhibited by typically achievable systemic concentrations.

Intermediate refers to organisms with MICs in the middle of the overall range of distribution of MICs. In some population distributions, there may be isolates that have no strains with fully susceptible MICs. For example, the action of penicillin or ampicillin against Enterococcus strains is such that serious infections with such strains would have to be treated synergistically with penicillin or ampicillin combined with an aminoglycoside, as already noted. For some organism–drug pairs, the moderately susceptible category represents a narrow range that is really intermediate or equivocal and cannot define clearly. This range is a buffer zone and allows for technical or biologic (plus or minus one dilution) variation in MIC reproducibility.

The resistant category encompasses organisms whose MICs indicate they are not readily amenable to treatment with that drug, although the tissue or drug concentration is readily attainable. It provides the most relevant clinical guidance of any of the interpretive categories and implies with reasonable assurance that normally a serious infection would not respond to treatment. However, it should be noted that, for the typical MICs interpreted as resistant, subinhibitory concentrations may produce a potentially beneficial clinical effect on some strains by modifying some pathogenicity traits such as rate of growth, ability to adhere, toxin production, or susceptibility to phagocytic action.

In the past, for selected multiresistant strains producing urinary tract infections, CLSI (NCCLS) documents defined a fourth category, “conditionally susceptible.” This category was utilized because of the high levels achieved by certain drugs and the occasional value of utilizing these drugs in uncomplicated urinary tract infections associated with multiresistant enteric bacteria with high MICs. This interpretive category is absent from current documents, and the intermediate category now encompasses this group.

A new term has emerged, “nonsusceptible,” which has two different interpretations. CLSI (NCCLS) uses this term for characterizing some organism–antimicrobial combinations where there are or very few resistant strains to define a resistant category. “Clinically nonsusceptible” refers to that population of microorganisms that qualify as either intermediate or resistant.

Because the interpretation of these guidelines cannot be arithmetically quantified and cannot be agreed upon with a great deal of reliability, this author has taken the view that the interpretive guidelines should clearly represent the extremes of the susceptibility spectrum (i.e., susceptible and resistant). Additionally, it should be recognized that the extreme categoric interpretations are relatively constant. The intermediate (or moderately susceptible) category is subject to greater fluctuations, due to periodic reevaluation. Interpretive standards for phenotypic screening and confirmation of ESBLs in Enterobacteriaceae (specifically E. coli, K. pneumoniae, and Klebsiella oxytoca) are presented in Table 3.39.

It must be emphasized that the breakpoints presented in Tables 3.32 and 3.33 and supported by CLSI are not necessarily the values used in all countries. Examples of breakpoints used in various countries are shown in Table 3.40(204). In addition to CLSI, there are working committees in other countries, notably France and Great Britain, that have been active in developing breakpoints based on experiences in their particular geographic regions (204).

At many institutions, considerable efforts have been made to educate physicians not trained as infectious disease specialists in the utilization of MICs. However, one frequently sees that in daily practice, MICs are poorly understood and interpreted. One type of error involves the choice of the wrong class of drug. For instance, an aminoglycoside might unnecessarily be used to treat a relatively benign infection with E. coli that is susceptible to multiple drugs merely because an MIC of 1 for amikacin appears to indicate greater susceptibility than the MIC of 2 for tetracycline or the MIC of 4 for ampicillin. Another type of error leads to a false categorical interpretation of the wrong drug within a class when the absolute values of the MICs are seemingly different. For these reasons, it is recommended that the appropriate interpretative category always be reported along with the numerical value of the MIC. In some clinical settings, there has been a tendency to report the ratio of the MIC to theoretical peak serum (or urine) levels for various dosages. The reports are sometimes computerized and are semiquantitated by a single plus sign or several plus signs, indicating the degree of susceptibility (or resistance). Such reports really are somewhat arbitrary and falsely quantitative. Although theoretically, such calculations are possible, they omit other PK parameters such as the possibility of protein binding (Should it be subtracted?), the use of peak or mean levels, and the use of trough levels in the numerator. What about renal or liver function tests, and in whom have the levels been determined? The patient under consideration has a specific disease and may not be compartmentalizing the drug in the typical way that the manufacturer’s product insert indicates for that antibiotic. Such specific pseudoquantitative reporting systems are based on oversimplified pharmacologic assumptions and may be misleading.

The general concept of therapeutic ratios is useful in teaching antimicrobial management. Thus, it may be appropriate and informative for laboratories to include, in their quarterly or annual reports, tables summarizing the standard interpretive breakpoints (as noted in Tables 3.32 and 3.33) and the ranges of mean serum levels. Additionally, this information can be placed at the back of a patient’s report. Lastly, since the bottom line is always cost, it is often necessary in many clinical settings for the laboratory, through the pharmacy and therapeutics committee, to supply clinicians with the average daily cost of an antibiotic, as determined at that institution, for treatment of severe to moderate infections. The clinicians can then use the laboratory susceptibility data along with the cost data to determine the most clinically efficacious and least expensive therapy.

Minimum Inhibitory Concentrations: Predictive Value of Clinical Outcome

Earlier, reference was made to the inoculum concentration, rate of replication, and phase as modifying factors in obtaining accurate, reproducible MIC results. Recently, it has been learned that microbial interactions in biofilms can confound the interpretation and predictive value of in vitro systems using homogenous bacterial populations. The model of the way bacteria reproduce and grow as free-swimming planktonic cells is not truly realized in nature. Planktonic cells attach to inanimate surfaces (e.g., plastic catheters or sand) or to the epithelial cells of an airway system. Their ability to detach or remain attached is dependent on several characteristics of the host system. Typically, the ability of clusters of bacteria to remain attached and form biofilms relates to availability of a foreign object (catheter) or dead tissue and to protection from host defenses and antibodies. Microbial biofilms are slow to develop, cause collateral damage to tissues, and are persistent. They are regarded as communities of bacteria that grow on surfaces, as opposed to microorganisms that are dispersed or free floating, and they complicate several medical problems, namely, periodontitis, osteomyelitis, infected catheters (pacemaker leads and urinary catheters), and cystic fibrosis.

It is expected that within communities the bacteria communicate. In the 1960s, marine microbiologists discovered that for Vibrio fischeri, the ability to produce light (bioluminescence) is dependent on a critical population size. The signal that turned on the light was genetically controlled by an autoinducer. Infectious disease specialists and microbiologists who study pathogenic mechanisms and virulence have learned that the changes in gene expression patterns in response to the host environment are a prerequisite for bacterial infection, and autoinducers and associated genes have been found in gram-positive and gram-negative bacteria. In studies of P. aeruginosa in cystic fibrosis patients, it was found that the prophylactic administration of macrolides that do not kill P. aeruginosa (e.g., azithromycin) can bring about clinical relief. This might be attributable to the activity of the drug on the normal oropharyngeal flora, thus minimizing the release of autoinducer-2 (AI-2) of P. aeruginosa. AI-2 is responsible for the induction of virulence genes for exotoxin and elastase (205). For glycopeptide intermediate-level resistant Staphylococcus aureus (GISA), the majority of infections originate on biomedical devices. The loss of accessory gene regulator (agr), a gene cluster comprising five different genes, involved in quorum sensing (the mechanism bacteria use to communicate with each other) has been suggested as contributing to their ability to produce biofilms. All VISA/GISA strains tested belong to agr group II and have defective agr function (206). In E. faecalis, the gene locus fsr has been identified as present in 70% of clinical isolates. It appears responsible for regulation of two virulence genes, for gelatinase and a serine protease (207). Homology between fsr and the agr gene of Staphylococcus has been noted (208).

The role of these virulence genes in diverse bacterial groups interacting with normal or commensal microflora and/or growing as biofilms poses challenging questions for medical microbiologists. What is the value of isolating a single offending species and determining the MICs of a variety of antimicrobial agents when, in situ, its role may be regulated by other bacterial populations within the community it resides in?

Given this background, what is the reliability of an MIC and its S/R interpretation in predicting the clinical outcome of antimicrobial therapy? Apart from pharmaceutical-sponsored studies of directed single-agent activity against a particular bacterial target or group, there is a paucity of reports on the relevance of in vitro bacterial susceptibility to the outcome of antimicrobial therapy.

In a retrospective study (209) of 510 patients who received antimicrobial therapy, 382 (75%) had susceptibility tests performed on at least one culture prior to the administration of antimicrobial therapy. Eighteen bacterial species (~75% due to gram-negative rods) were recovered from 298 patients, and of these patients, 271 (91%) received antimicrobial therapy to which the organisms were susceptible and 219 (81%) improved. Of the 271 patients who received therapy to which the bacteria were resistant, 3% demonstrated improvement and 82% did not improve. This study clearly shows the value of selecting therapy according to in vitro susceptibility test results. Similarly, a prospective observational study of 2,634 septic patients showed that “adequate antibiotic treatment” defined on the basis of in vitro susceptibility of an isolated microorganism (at least five species were identified) and/or initiation of antibiotic treatment between 24 hours before and 72 hours after study enrollment resulted in a 10% decrease (33% vs. 43%) in mortality (210). In contrast, an international prospective observational study (211) of 844 hospitalized patients with blood cultures positive for S. pneumoniae reported that “discordant therapy” (inactive in vitro susceptibility with penicillins, cefotaxime, and ceftriaxone but not cefuroxime), as compared with “concordant antibiotic therapy” (i.e., receipt of a single antibiotic with in vitro activity against S. pneumoniae), did not result in a higher mortality rate. Similarly, the time required for defervescence and the frequency of suppurative complication did not result in a higher mortality rate. The conclusion from these data is that β-lactam antibiotics could prove useful for pneumococcal bacteremia regardless of in vitro susceptibility as defined by CLSI (NCCLS) breakpoints.

Research on acute exacerbation of chronic bronchitis (AECB) includes several studies that compared the activity of quinolones and macrolides against H. influenzae. There are differences in the impact of compounds of these two types. Apart from the direct antibacterial PD properties of these agents, the macrolides appear to possess an immunomodulatory effect that prevents recurrence of infection. A conclusion of Martinez in cases of AECB is that in vitro antimicrobial resistance is of unclear significance (212,213).

Against nosocomial pneumonia, antimicrobial agents are the mainstay of pharmacologic measures. It is worth noting that in these studies gram-negative bacteria were the most frequently isolated microorganisms. Several studies examined the mortality rate in association with inadequate antimicrobial therapy. When antibiotic therapy was “appropriate” (i.e., the infective organisms were susceptible), there was a 60% decrease in mortality from 90% (inappropriate) therapy to 30% (appropriate) (214–216).

Prior to the publication of the Yu et al.’s (211) study cited earlier, Rex and Pfaller (217) reviewed multiple reports examining the correlation of therapeutic outcome with in vitro susceptibility. They proposed the “90–60 rule,” which states that infections that are due to susceptible isolates respond to appropriate therapy about 90% of the time, whereas infections that are due to resistant isolates or are treated inappropriately respond about 60% of the time.

MIC values estimate the bacteriostatic or inhibitory activity of antimicrobial agents. An MIC, when determined according to the standards and references detailed, is a reproducible parameter for a given antimicrobial agent against a variety of rapidly growing pathogens. In clinical practice, the MIC usually suffices for guiding chemotherapy. The success of in vivo antimicrobial action depends to a large extent on the host’s defense mechanisms, which ultimately sequester and kill the microorganisms that have been reduced by the bacteriostatic action of the chemotherapeutic agent. The main body of medical microbiology, clinical pharmacology, and infectious disease literature utilizes MIC data in studying the effects of antimicrobials and in establishing criteria for application in therapy.

For antimicrobial agents that possess bactericidal action (mainly aminoglycosides and β-lactams), it is sometimes necessary to perform additional quantitative assessments of the killing effect on a given offending microorganism. The parameter known as the MBC can be determined in several ways:

 1.  By estimating the MBC as a result of the MIC for an infecting organism.

 2.  By estimating the titer of serum of a patient receiving antimicrobial therapy that kills the infecting organism after fences (i.e., the serum bactericidal titer or test [SBT]).

 3.  By determining the number of surviving bacteria in a fixed concentration of the drug using the average obtainable blood level at defined time intervals (i.e., the killing curve).

The assessment of bactericidal activity, although methodologically feasible, is fraught with microbiologic phenomena and technical problems requiring consummate understanding on the part of those who consider their application and those who execute the assays.

Microbiologic Factors

Since the interactions of bacteria and antimicrobial agents began to be gauged, investigators have observed unusual and complex phenomena that remain incompletely understood in the modern molecular era. One such phenomenon is known as the paradoxical effect (or the Eagle phenomenon) (55). Discovered in the early days of penicillin therapy, the paradoxical effect manifests as the puzzling appearance of increasing numbers of bacterial survivors at concentrations higher than the MBC. Since its discovery, it has been observed for several species of bacteria and for antimicrobial agents other than the β-lactam group and is believed to be the result of interference with protein synthesis of the organism by higher concentrations of the β-lactam. The paradox occurs when the proportion of surviving cells increases significantly even as the concentration of antimicrobial agent increases beyond the MBC. It has been theorized that the high concentration of antimicrobial agent inhibits protein synthesis to a degree that prevents the growth necessary for expression of the lethal effect of the drug. No therapeutic implications appear associated with this effect.

A second complexity involving incomplete killing has been referred to as the persistence phenomenon and relates to the small proportion (usually less than 0.1%) of the inoculum cells that persist (survive) despite the lethal activity of antimicrobial agents. Again, this is especially common with β-lactam agents. If the persisters are subcultured and retested, they appear as susceptible to the effects of the antimicrobial agent as the original isolate, and no greater proportion of cells persist. Persisters have been considered to be metabolically inactive forms that were not actively growing at the time of the interaction of the drug with the inoculum and consequently were not killed by the β-lactam compound.

Of the several mechanisms by which bacteria seemingly evade the killing effect of antimicrobial agents, perhaps the least understood is that of tolerance. The term was coined in 1970 by Tomasz et al. (218) to describe the atypical in vitro response of pneumococcal strains to penicillin. This response was later recognized as genotypic tolerance. The more common and perhaps more clinically relevant type of tolerance, phenotypic tolerance, was described earlier, in 1942, by Hobby et al. (219).

Reports related to the phenomenon of tolerance have accumulated rapidly over the past few years, and several reviews deal with it (220–223). Operationally, tolerance can be defined as the ability of bacteria to grow in the presence of high concentrations of antimicrobials, so that the killing action of the drug is avoided but the MIC remains the same. In Tomasz’s (218,226) original observation, typical pneumococcal isolates were quickly lysed and lost viability at penicillin concentrations greater than the MIC. However, the selected tolerant strain was not lysed and lost viability at a significantly reduced rate. Recently, it has been suggested that four different mechanisms may contribute to the ability of organisms to survive (or survive at a higher rate) during treatment with penicillin or other cell wall–directed compounds. Tuomanen et al. (223) proposed the term survivor mutation and attributed the mechanism of survival to ancillary bactericidal and lytic processes. Moreover, they cautioned against using the term tolerant to describe all of these isolates. The relationship between the killing rates in different bacterial populations and the MBC is depicted in Figure 3.17.

Tolerance in the main has been associated with lactam agents and has been reported for a number of genera, including Streptococcus, Staphylococcus, Listeria, Lactobacillus, and Clostridium. A growing debate revolves around the criteria used to define tolerance. The generally acceptable definition has been a ratio of the MBC to the MIC of more than 32, as originally defined by Sabath et al. (224). However, differences among investigators have led to an assortment of values ranging from 8 to 32 to 100 (221). A more precise approach for detecting tolerant strains is to use quantitative killing curve methods. More than 20 bacterial species recovered from clinical material have been implicated as tolerant strains. However, because of a lack of consensus, the true incidence of tolerance among clinical isolates remains to be determined. The problem is muddied by the variable application of adequate bacteriologic techniques, the variable definition of tolerance, and the lack of suitable reference strains.

To study the effects of antimicrobials on bacteria with decreased or arrested division rates, two new parameters have been suggested for assessing the efficacy of killing. One of them, the MnBC, is analogous to the MBC but applicable to conditions (phenotypic) of slower growth; it is defined as the concentration of drug that achieves a 1-log killing in 24 hours of cells starved for 10 minutes before the addition of antibiotic (223).

Whatever the terms and categories used to describe these events, it is necessary to recognize that the organisms involved may be present in various clinical situations. Although it is generally accepted that rapidly growing and dividing organisms are more susceptible to the inhibiting effects of cell wall–directed antimicrobial agents, it is recognized that rapidly growing bacteria flourish under broth-related clinical conditions like bacteremia. In other clinical conditions, specifically osteomyelitis, it has been demonstrated that microorganisms divide at a much reduced rate and thus would be less susceptible to the effects of antimicrobial agents. A perceived problem in dealing with this clinical dilemma is how to select antimicrobial agents that would be well targeted in arrested growth situations (222). It is of utmost interest and importance to determine whether these isolates are the result of technical manipulation in the laboratory or are indicative of a real clinical phenomenon (Fig. 3.18).

The concept of tolerance is derived from the bactericidal mode of action of β-lactam antibiotics. β-Lactam compounds are bactericidal because they inhibit bacterial cell wall synthesis. Basically, their mode of action is to interfere with the transpeptidation process that links the individual peptidoglycan components of the bacterial cell wall to each other (35,68,225–227). β-Lactams bind to and inactivate specific targets on the inner surface of the bacterial cell membranes. These targets are referred to as the PBPs (160,226–229). The PBPs are enzymes—transpeptidases, carboxypeptidases, and endopeptidases—involved in the terminal stages of assembling the bacterial cell wall and maintaining the structure of the cell wall during growth and division (230,277). β-Lactam compounds have different attachment sites and binding capacities for the various PBPs and, depending on the specific PBP bound, have different effects on bacteria (231–233). The inactivation of some PBPs (PBPs 1A, 1BS, 2, and 3) contributes to bacterial death. In contrast, other PBPs (PBPs 4, 5, and 6) are not essential for bacterial viability, and their inactivation by β-lactam molecules is not lethal for the bacteria (234,279). Currently, it is theorized that β-lactams binding to PBPs inactivate endogenous inhibitors of bacterial autolysins. The autolysins can then disrupt covalent bonds in the bacterial cell wall and cause bacteriolysis. The growth of certain organisms that lack autolysins can be inhibited by β-lactam antibiotics, but such organisms are not killed, thus leading to the phenomenon of bacterial tolerance (44).

Clinical Relevance and Applications

In some clinical situations, both microbiologic and clinical data have been accumulated that suggest that MBC determinations or other bactericidal assays may be interpretable and relevant. These have been referred to earlier and include directed activity against Enterococcus strains associated with endocarditis (235) and occasional cases of bacterial endocarditis (236) caused by other organisms that may not be fully susceptible, particularly Pseudomonasspecies, MRSA, various species of coagulase-negative Staphylococcus, and S. aureus. The MBC may also be of value in the management of osteomyelitis (237) and bacteremic infections in granulocytopenic patients (238).

Technical Factors

The application of the MBC (which is founded on the MIC and its derivative, the MBC/MIC ratio) to determine the microbicidal activity of antimicrobial agents has been questioned because the end points are based on arbitrary definitions and are often poorly reproducible. Even within a particular methodology, such as macrodilution versus microdilution procedures (239), there is great biologic variability and different end points may be obtained (Table 3.41).

Dilution methods suffer from several technical problems such as antibiotic carryover, bacteria adhering to the surface of the test vessel, and variations in the medium and growth phase of the inoculum, all of which affect the CFU recovered from subculture plates (240,241) (see “Minimal Bactericidal Concentration Procedure”). One step in performing the assay is critical because of an unusual condition that occurs at the surface (meniscus) of the assay container. At the medium interface, viable organisms flourish, perhaps to escape the potential lethal action of the antimicrobial agent. This potential for error is diminished by mixing and reincubating (see “Minimal Bactericidal Concentration” steps 13 and 14).

As for MIC assays, the recommended broth is MHB supplemented with Ca²⁺ and Mg²⁺ for testing P. aeruginosa. NaCl (2% final concentration) should be added for testing S. aureus. MHB can be used with human serum (HS) in a 1:1 ratio. The use of HS depends on the antimicrobial agent to be tested (and its potential for protein binding), the organism, and the bactericidal test executed. The selection of medium may be altered for research needs in order to grow fastidious microorganisms. For the serum bacterial titer, a 1:1 combination of MHB and HS is the recommended medium (242).

Adherence to the details outlined in the preceding sections with respect to inoculum size, strain storage, growth phase, assay medium, cation content, and incubation duration and temperature must be strict.

Whereas it is anticipated that for bactericidal drugs the MIC and MBC would be similar, it is accepted that for bacteriostatic drugs the MBC could be several dilutions greater. A procedural problem related to the MBC is the definition of the end point as it relates to the number of survivors that remain in the population, because 100% elimination is an impossible goal to achieve (241,243). Several investigators have utilized the definition of a reduction of the number of bacteria present in the inoculum to 99.99% of the original population. This represents a 10⁻³ (−log 10³) reduction and is somewhat arbitrary, since there is no convincing evidence that a 99.0% or 98.0% reduction actually portrays the outcome and is more clinically relevant. In the United States, CLSI (NCCLS) (180) has written a proposed standard for these tests in which a 1,000-fold reduction of the original inoculum is used as a conventional standard. When the ratio of MBC to MIC is 32 or greater for a given bacterium–antimicrobial combination, the organism is said to be tolerant to the action of the antimicrobial and it is questionable whether a favorable clinical response or outcome can be achieved. The popular definition of tolerance (MBC/MIC greater than 32) has little scientific basis and results in organisms whose response to penicillin is close to the 99.9% definition of kill being artificially divided into susceptible and tolerant categories (240) (Fig. 3.19). In view of the lack of reproducibility of conventional MBC methods, the definition should be treated with caution.

Minimum Bactericidal Concentration

The important studies by Taylor et al. (244) on the MBC determination for staphylococci have shown that much of the variability in the assay results is due to procedural details. In the United States, most of the contention regarding the assay revolves around the issue of whether to use the macrodilution or microdilution method. In an extensive investigation, James (245) compared four methods for determining the MIC and MBC of penicillin against S. viridans. In his studies, the macrodilution and microdilution methods were compared, along with the membrane and gradient plate methods. The author studied 28 strains of S. viridans streptococci by these methods and determined the MICs and MBCs. From these data, he calculated the mean error span shown in Table 3.42. Note that conventional MIC methods are expected to yield reproducibility between tests within one doubling dilution, which equates to a mean error span of less than 0.5 tube errors. If this criterion is applied, then the macrodilution, membrane, and gradient methods gave acceptable reproducibility, whereas the microdilution method did not and therefore should not be used for the determination of the MIC, at least with S. viridans streptococci. When one applies the same criterion to MBC determinations, only the gradient method gave acceptable reproducibility, and it was the most reliable method for predicting penicillin tolerance. James (245) concluded from his studies that the mean error spans for all methods were acceptable, except for the microdilution technique. However, correct and reproducible results were obtained for all control organisms by the gradient method and, to a lesser extent, the membrane method. His study clearly indicated that the microdilution results were unacceptable.

In deciding on a method to adopt for determining the MBC, it is thus best, as shown by the work of James (245), to avoid the microdilution method. Because many reference and clinical laboratories may not have accumulated experience with the gradient or membrane methods, the macrodilution method is the logical choice. It is described here in detail, as adapted from Schoenknecht et al. (241).

Minimum Bactericidal Concentration Procedure

Minimum Inhibitory Concentration

 1.  Subculture organisms onto appropriate medium (usually a blood agar plate) and incubate overnight at 35°C.

 2.  Inoculate a tube containing 3 mL of saline or MHB with five or more colonies from the overnight plate to achieve a turbidity equivalent to a no. 1 McFarland standard (approximately 10⁸ organisms/mL).

 3.  Transfer 0.1 mL of turbid inoculum (patient’s pathogen) into 10 mL of MHB or other appropriate broth. Incubate in a shaking water-bath or equivalent at 35°C until turbid. This corresponds to an end point between a no. 1 McFarland standard and an overnight suspension and requires 5 or 6 hours for rapid growers.

 4.  Inoculate the standard control organism (E. coli, S. aureus, etc.) into 3 mL of broth and incubate (without shaking) at 35°C until turbid.

 5.  Prepare twofold serial dilutions of the antibiotic in 2 mL of MHB (total volume per acid-washed borosilicate glass tube); 16 × 100-mm glass tubes with loose-fitting metal caps are preferred.

 6.  Standardize the inocula (patient’s organism and control organism) to equal a 0.5 McFarland turbidity standard (approximately 5 × 10⁷ organisms/mL) in 3 mL of saline or broth.

 7.  Dilute adjusted inocula 1:10 (0.2 mL in 1.8 mL of MHB or appropriate substitute). This equals about 5 × 10⁶ organisms/mL.

 8.  Dispense, using an Eppendorf or equivalent pipette, 100 µL of diluted inoculum into tubes containing serial dilutions of the antibiotic. To inoculate, insert the pipette tip well under the surface of the antibiotic-containing broth. Avoid any contact between the tip and the walls of the tube. Rinse the tip five times in solution. The same tip may be used throughout the test if inoculating from lowest to highest concentration of antibiotic. The final inoculum size is approximately 2.5 × 10⁵ organisms/mL.

 9.  Incubate for 20 hours at 35°C.

10.  From the 1:10 dilution of the 0.5 McFarland-adjusted inoculum, which should be about 5 × 10⁶ organisms/mL (step 7), dilute serially 1:10 in MHB four times to achieve a final inoculum of 5 × 10²organisms/mL as follows: 0.2 mL (5 × 10⁶ CFU/mL) + 1.8 mL MHB; 0.2 mL (5 × 10⁶ CFU/mL) + 1.8 mL MHB 5 × 10⁵ CFU/mL; 0.2 mL (5 × 10⁶ CFU/mL) + 1.8 mL MHB 5 × 10⁴ CFU/mL; 0.2 mL (5 × 10⁴ CFU/mL) + 1.8 mL MHB 5 × 10³ CFU/mL; 0.2 mL (5 × 10³ CFU/mL) + 1.8 mL MHB 5 ×10² CFU/mL.

11.  Aliquot 0.1 mL of this suspension, dispense either into a tube of melted agar for the preparation of pour plates or onto an appropriate agar plate (e.g., blood agar) and distribute evenly using sterile bent glass rods. This procedure should be done in duplicate. Incubate overnight at 35°C.

12.  Observe and record MIC of control organisms.

13.  For patient’s sample only, vigorously vortex-mix tubes without visible growth for 15 seconds and reincubate for an additional 4 hours.

14.  Vortex-mix again and sample tubes for MBC determination; spread 100-µL samples across the surface of dried TS agar plates with sterile bent glass rods.

15.  Record patient’s MIC.

Minimum Bactericidal Concentration

16.  Incubate plates overnight at 35°C for the MBC test.

17.  After 1 day (or 2 days), count the number of colonies per plate from the original inoculum plates or pour plates and average. Determine a colony count that represents 0.1% of the original inoculum (i.e., 99.9% reduction).

18.  Count colonies from MBC plates. Any number equal to or less than the determined colony count from step 17 is considered as a 99.9% kill or bactericidal result.

When counting the number of colonies to determine the plate average, the mean is referred to as N. Apply the formulato determine the upper limit of a colony count that represents 0.1% of the original inoculum (approximately 95% confidence limits). Therefore, any colony counts from MBC plates equal to or less than the determined inoculum colony count upper limit are considered 99.9% kill or bactericidal results.

Alternatively, rejection values can be determined from a chart that takes into account the final inoculum size, dual sampling, pipetting error, and the Poisson distribution of sample responses (Table 3.43).

Hacek et al. (246), recognizing the potential variability and complexity imposed by this assay, proposed a modified scheme. Their modified bactericidal testing protocol includes omitting serum supplementation, incubation without agitation, running tests in duplicate with a reduced number of dilutions (six instead of nine), extending the incubation interval for 24 hours to resolve discrepancies, using single 0.1-mL aliquots, and adopting an alternate end point calculation. The authors found a 91% agreement between the standard and their modified protocol and suggest the alternative procedure is practical for the clinical laboratory.

Serum Bactericidal Titer/Test

Clinical Relevance and Applications

The seminal work on and application of this test were done by Schlichter and MacLean (247). In its original form, the test determined bacteriostatic activity; it was later modified to include bactericidal activity. However, in the nearly 50 years that the test has been available and used, there has been no clear, universally accepted criteria for its application. Critical reviews of the SBT have not found the test clinically useful and have stressed the need for standardization of the methodology (237,248). CLSI (NCCLS) has developed a proposed guideline (249), and in a thoughtful review, Stratton (91) addressed the specific application and clinical relevance of the test.

Assessing the antimicrobial activity in a patient’s serum during treatment by using the offending organisms isolated from the patient as the test strain would appear to be the most logical approach to evaluating and monitoring chemotherapy. The SBT measures the combined effects of absorption and elimination of the antimicrobial agent; its potential binding to serum proteins; the effect of metabolic congeners of the parent compound against the microorganisms; and, if dual antimicrobial therapy is administered, the effects of drug interactions, including synergistic, additive, and antagonistic effects.

As with the determination as MBC, the SBT can be helpful in monitoring the treatment of bacterial endocarditis (235,236), bacteremia in patients with cancer (150), osteomyelitis or septic arthritis (237,250), and bacterial meningitis (251).

As an experimental tool, the SBT has been used in evaluating new drugs and drug combinations and detecting antimicrobial potency in infected body fluids other than blood. When the SBT has been used for drug evaluation (in humans or animals), analysis of the results can be aided by using the titer to measure the area under the bactericidal curve (AUBC) (252).

Although the SBT approach seems to be a logical integration of physiologic (pharmacokinetic) and in vitro susceptibility, this test, as well as other bacterium-antimicrobial assays, fails to evaluate the cellular and humoral defenses of the host, the site and severity of the infection, the quantity of bacteria present and their virulence, and the continually changing concentration of the antimicrobial agent in the host.

Serum Bactericidal Titer/Test Methodology

As with the MBC procedures, many variations of this test exist. The same care and attention to details and materials used in executing the MBC procedures should be used here. The method that follows is adapted from several sources (241,243,249).

Collection of Patient Serum

Inherent in this procedure is the use of the patient’s serum to represent the physiologic concentration of the antimicrobial agent. For this purpose, a peak level and a trough level are generally obtained. The peak level is considered the level obtained 30 to 45 minutes after an intravenous infusion, 60 minutes after an intramuscular infusion, or 90 minutes after an oral dose. The peak level is obtained 30 to 60 minutes after the drug is absorbed and distributed. The trough level is the level that is considered to occur 30 minutes prior to the following dose.

After the specimen has been collected, it should be transported to the laboratory properly and promptly, to separate the blood and serum. Serum collected from the specimen should be frozen if a delay of more than 2 hours in performing the SBT is anticipated. Ideally, trough and peak serum specimens should be pair-matched rather than collected on different days of therapy.

Patient Organism

The SBT must be anticipated so that the patient’s clinical isolate can be saved. Because it is the practice of many clinical laboratories to save bacteremic isolates as well as isolates from spinal and other body fluids, these can be retrieved early. On the other hand, if it is anticipated that the isolate will be needed, this should be kept frozen at 70°C in a TS broth or in a cryoprotectant medium (e.g., glycerol).

Serum Bactericidal Titer/Test Procedure

 1.  Serially dilute the patient’s serum twofold at least 1:64, using MHB as diluent. The final volume per tube should be 1 mL.

 2.  Add the patient’s organism to a 0.1-mL volume of a carefully prepared inoculum (see steps 10 and 11 of the MBC procedure).

 3.  Prepare pour plates for establishing original inoculum CFU counts (see steps 7 through 11 of the MBC procedure).

 4.  Include a growth control tube containing MHB and inoculum but no serum.

 5.  Incubate for 18 to 24 hours.

 6.  Subculture for the 99.9% bactericidal end point, as described for the MBC procedure.

Interpretation

The interpretation of end points is controversial. The CLSI (NCCLS) guidelines (249) offer the following: peak titer of 1:2 and trough titer of 1:2, interpretation of inadequate; peak titer of 1:4 to 1:16 and trough titer of 1:4, interpretation of intermediate; and peak titer of 1:32 and trough titer of 1:8, interpretation of adequate. According to Stratton (237), in orthopedic infections, a titer of 1:8 is prognostic of cure, although higher titers do not necessarily preclude positive outcomes.

Killing Curve

The killing curve or killing rate is represented by a plot of the number of survivors in the host after administration of a typical therapeutic regimen. It has been used to evaluate and compare new drugs and to study differences and changes in the antimicrobial susceptibility of clinically important bacterial isolates. These determinations are rarely used for guiding chemotherapy; they are mainly applied in experimental situations to animal models and are generally used to assess classes of drugs. One concentration of antibiotic is tested, usually that which is representative of an average level obtainable during therapy. At periodic intervals, usually at 0, 4, 12, and 24 hours of incubation, colony counts are performed and charted on semilogarithmic paper, with the survivor colony count on the ordinate (logarithmic scale) and time on the abscissa (arithmetic scale) (Fig. 3.20). For example, when one compares the β-lactam antimicrobial agents with the aminoglycosides, the former are characterized by slower, dose-dependent initial bactericidal activity. The extent of the bactericidal action is related to the time during which the serum level exceeds the MIC. If this level falls below the MIC, there is immediate regrowth of the microorganisms. In contrast, aminoglycosides demonstrate rapid, dose-dependent initial bactericidal activity, followed by a bacteriostatic phase that can last several hours after the serum concentration falls below the MIC.

The kinetics of antimicrobial activity, as described, provides a theoretical basis for dosing frequency and depends on the pathogen and the antibiotic used. Only one antibiotic concentration, representing the average obtainable blood level, or a limited number of concentrations, representing multiples or fractions of the blood level, are used. In the plot in Figure 3.20, the aminoglycoside is shown to be a drug with concentration-dependent bactericidal activity. At increasing drug concentrations, there is an increase in the magnitude and rate of killing. β-lactam compounds (penicillin in Fig. 3.20) demonstrate little concentration-dependent bactericidal activity. At least for two antimicrobial agents, ampicillin and ciprofloxacin, their activities (expressed as killing rates) were similar in MHB and human urine, as judged by an in vitro PD model (40).

Although the protocol outlined in the following text relates to evaluating the lethal activity of individual antimicrobial agents, agents combined for their potential synergistism can also be tested (see Chapter 10).

More recently, modified killing curve investigations have been used to elucidate anticipated multiple resistance mechanisms expressed by S. pneumoniae and Mycobacteria species against fluoroquinolone compounds (253,254). In this variation of the killing curve procedure, large numbers of microorganisms (10⁸ to 10¹⁰) are plated on increasing concentrations of antibiotic—in this case, fluoroquinolones—and the fraction of cells that can be recovered are then treated as CFU. The fraction that survive or grow are plotted against the concentration (i.e., the MIC). As can be seen in Figure 3.21, the first dip in the plot occurs at the MIC; a second inflection point in mutant recovery occurs at a concentration the authors refer to as the mutant prevention concentration (MPC), the concentration required to block first-step mutants. This approach has proved valuable in studying the design of the fluoroquinolones in order to anticipate the possible development of stepwise mutants.

Killing Curve Procedure

 1.  Prepare inoculum of approximately 5 × 10⁷ CFU/mL, as in steps 1 through 3 and 5 and 6 of the MBC procedure.

 2.  Inoculate according to step 8 of the MBC procedure. The final concentration should be 5 × 10⁵ CFU/mL for each tube.

 3.  Immediately vortex-mix for 15 seconds the zero growth tube and dispense 0.1 mL into a tube of melted agar for the preparation of agar pour plates. Incubate at 35°C for 18 to 24 hours.

 4.  Similarly, for each subsequent time interval, prepare pour plates.

 5.  For testing at 24 hours, vortex-mix the last tube at 20 hours and reincubate for an additional 4 hours.

 6.  At 24 hours, vortex-mix again and prepare pour plates.

Serum Bactericidal Rate

The time-kill curve evaluates the rate at which single or multiple drugs kill bacteria. The serum bactericidal rate represents the rate of serum killing; the method used integrates in vitro activity and the in vivo pharmacokinetics of antimicrobial drugs and therefore reflects very accurately what happens in the host.

In a novel approach to the application of the serum bactericidal rate, Barriere et al. (252) measured the AUBC to determine the total synergistic bactericidal activity. When the calculated value of the AUBC for a combination was greater than the sum of the values for the individual drugs, the combination was judged to be synergistic.

A therapeutic role for the use of killing curves was shown by Small and Chambers (255). They studied the clinical failure of vancomycin when treating intravenous drug users with S. aureus endocarditis. In vitro studies with 10 strains of S. aureus recovered from this patient group demonstrated that, at four times the MIC, vancomycin was less rapidly bactericidal than nafcillin (Fig. 3.22). At 4 hours after incubation, the mean decreases in bacterial counts for vancomycin and the β-lactam were similar. However, after 24 hours, the mean count for vancomycin had not significantly changed from the 4-hour titration, although there was a significant reduction (2.8 log₁₀ CFU/mL) for nafcillin. It was on this basis that the authors explained relapses and persistent bacteremia in the vancomycin-treated group and concluded that vancomycin was less effective than nafcillin for treating this infection.

Cytokines represent a broad group of soluble mediators of cell-to-cell communication; the group includes interleukins, interferons, and colony-stimulating factors (172,272). Cytokine molecules are mediators of specific and nonspecific host defense responses. As such, they play a critical role in effector mechanisms that eliminate foreign antigens such as microorganisms. Advances in our understanding of cytokines have focused on two primary biologic activities: the regulation of inflammation with effective immune responses to pathogen invasions and a wide array of hemopoietic activities that serve to modulate the growth of immune cells. Table 3.44 lists some cytokines and their biologic effects. It is with the understanding that cytokines may play a supportive or synergistic role in conjuction with antimicrobial agents that they are included and discussed here. The interferons have demonstrated significant activity in different types of infectious disease, which has resulted in several FDA-approved indications (256). Interferon- has shown antiviral activity, and some interleukins are known to demonstrate protection against intracellular pathogens. Interleukin-1 receptor antagonist serves as an active mediator during severe bacterial infection or septic shock. As recombinant forms of these compounds become available, no doubt clinical trials will be attempted to determine effectiveness.

The cytokines can be readily assayed in human biologic fluids by an enzyme-linked immunosorbent assay–type sandwich immunoassay that recognizes both natural human and E. coli–derived human cytokines. A typical standard curve from a commercial kit assay is shown in Figure 3.23.

Laboratory assessments of antimicrobial activity will no doubt attempt to more broadly incorporate microbiologic and pharmacologic attributes of antibiotics. Advances in PK/PD modeling have made significant contributions to establishing breakpoint determinations that are more clinically relevant. However, the application to direct patient care in real time has not been fully realized. As currently conceived, some of the approaches that use computer-simulated models are too tedious to execute as part of actionable health care. Aspects of these interactive models as suggested have not as yet been incorporated into “apps” available on mobile devices to be used at the patient’s bedside to help in the delivery of rational, safe, and effective therapy (257). As new pathogens emerge, many associated with resistance mechanisms that threaten to limit the effective therapeutic use of antimicrobial agents, clinical laboratories will be called upon to move rapidly and accurately to screen for the presence of resistant members. Toward this end, methods for detecting genetic markers of resistance such as those discussed in this volume will be used. Additionally, the use of multiple probes directed toward more than one gene target (multiplexing) need to be designed as panels to detect the several antimicrobial-resistant phenotypes. Lastly, as the pipeline for the development of new antimicrobial compounds maintains its constricted development, pharmaceutical companies may consider developing additional innovative, antiinflammatory, antisepsis medicines. At that time, clinical laboratories will have the capability to titrate inflammatory cytokines and their immunotherapeutic targets.

  1.  Fleming A. On the antibacterial action of cultures of a penicillium with special reference to their use in the isolation of B. influenzae. Br J Exp Pathol 1929;10:226–236.

  2.  Rammelkamp CH, Maxon T. Resistance of Staphylococcus aureus to the action of penicillin. Proc Soc Exp Biol Med 1942;51:386–389.

  3.  Foster JW, Woodruff HB. Microbiological aspects of penicillin. J Bacteriol 1943;46:187–202.

  4.  Vincent JG, Vincent HW. Filter paper modification of the Oxford cup penicillin determination. Proc Soc Exp Biol Med 1944;55:162–164.

  5.  World Health Organization. Standardization of methods for conducting microbic sensitivity tests: second report of the Expert Committee on Antibiotics. World Health Organization technical reports series, no. 210. New York: World Health Organization, 1961.

  6.  Ericsson HM, Sherris JC. Antibiotic sensitivity testing: report of an international collaborative study. Acta Pathol Microbiol Scand 1971;217(Suppl B):390.

  7.  Bauer AW, Kirby WMM, Sherris JC, et al. Antibiotic susceptibility testing by a standardized single disk method. Am J Clin Pathol 1966;45:493–496.

  8.  Federal Register. Rules and regulations: antibiotic susceptibility disks: correction. Federal Register 1973;38:2576.

  9.  Federal Register. Rules and regulations: antibiotic susceptibility discs. Federal Register 1972;37:20525–20529.

 10.  National Committee for Clinical Laboratory Standards. Methods for dilution antimicrobial susceptibility tests for bacteria that grow aerobically: approved standard. 6th ed. Wayne, PA: National Committee for Clinical Laboratory Standards. 2003. NCCLS document M7-A6.

 11.  National Committee for Clinical Laboratory Standards. Performance standards for antimicrobial disk susceptibility tests: approved standard. 8th ed. Wayne, PA: National Committee for Clinical Laboratory Standards, 2003. NCCLS document M2-A8.

 12.  D’Amato RF, Hochstein L, Vernaleg JR, et al. Evaluation of the BIOGRAM antimicrobial susceptibility test system. J Clin Microbiol 1985;22:793–798.

 13.  Matsen J, Koepcke M, Quie P. Evaluation of the Bauer-Kirby-Sherris-Turck and single-disk diffusion methods of antibiotic susceptibility testing. Antimicrob Agents Chemother 1970;9:445–453.

 14.  Amsterdam D. Principles of antibiotic testing in the laboratory. In: Balows A, Hausler WJ Jr, Lennette EH, eds. Laboratory diagnosis of infectious disease: principles and practice. Vol 1. New York: Springer-Verlag, 1988:22–38.

 15.  Lorian V. In vitro simulation of in vivo conditions: physical state of the culture medium. J Clin Microbiol 1989;27:2403–2406.

 16.  Appelbaum PC. Worldwide development of antibiotic resistance in pneumococci. Eur J Clin Microbiol 1987;6:367–377.

 17.  LaForce FM. Pneumonia3—41987: new developments. Eur J Clin Microbiol 1987;6:613–617.

 18.  Woolfrey BF, Fox JMK, Lally RT, et al. Broth microdilution testing of Pseudomonas aeruginosa and aminoglycosides: need for employing dilutions differing by small arithmetic increments. J Clin Microbiol 1982;16:663–667.

 19.  Moellering RC, Wennersten C, Medrek T, et al. Prevalence of high level resistance to aminoglycosides in clinical isolates of enterococci. Antimicrob Agents Chemother 1970;10:335–340.

 20.  Brumfitt W, Percival A. Laboratory control of antibiotic therapy in urinary tract infection. Ann NY Acad Sci 1967;145:329–343.

 21.  Trissel LA. Handbook on injectable drugs. Bethesda, MD: American Society of Hospital Pharmacists, 1983.

 22.  Barry AL, Lasner RA. Inhibition of bacterial growth by the nitrofurantoin solvent dimethylformamide. Antimicrob Agents Chemother 1976;9:549–550.

 23.  Murray PR, Niles AC. Effect of filtration on antimicrobial solutions. Antimicrob Agents Chemother 1981;20:686–687.

 24.  Barry AL. The antimicrobic susceptibility test: principles and practices. Philadelphia: Lea & Febiger, 1976.

 25.  Grove DC, Randall WA. Assay methods of antibiotics: methods for the determination of bacterial susceptibility to antibiotics. New York: Medical Encyclopedia, 1955.

 26.  Bauer AW, Sherris JC. The determination of sulfonamide susceptibility of bacteria. Chemotherapia 1964;9:1–19.

 27.  Binford JS Jr, Binford LF, Adler P. A semiautomated microcalorimetric method of antibiotic sensitivity testing. Am J Clin Pathol 1973;59:86–94.

 28.  Barry AL, Jones RN, Gavan TL. Evaluation of the micromedia system for quantitative antimicrobial drug susceptibility testing: a collaborative study. Antimicrob Agents Chemother 1978;13:61–69.

 29.  Gavan TL, Barry AL. Microdilution test procedures. In: Lennette EH, Spaulding EH, Truant JP, eds. Manual of clinical microbiology. 3rd ed. Washington, DC: American Society for Microbiology, 1980:359–362.

 30.  Gavan TL, Jones RN, Barry AL. Evaluation of the Sensititre system for quantitative antimicrobial drug susceptibility testing: a collaborative study. Antimicrob Agents Chemother 1980;17:464–469.

 31.  MacLowry JD, Jaqua MJ, Selepak ST. Detailed methodology and implementation of a semi-automated serial dilution microtechnique for antimicrobial susceptibility testing. Appl Microbiol 1970;20:46–53.

 32.  Washington JA II, Sutter VL. Dilution susceptibility test: agar and macro-broth dilution procedures. In: Lennette EH, Spaulding EH, Truant JI, eds. Manual of clinical microbiology. Washington, DC: American Society for Microbiology, 1980:453–458.

 33.  Mueller JH, Hinton J. A protein-free medium for the primary isolation of the gonococcus and meningococcus. Proc Soc Exp Biol Med 1941;48:330–333.

 34.  Chow BF, McKee CM. Interaction between crystalline penicillin and human plasma proteins. Science 1945;101:67–68.

 35.  Tilton RC, Newberg L. Standardization of the microdilution susceptibility test. In: Balows A, ed. Current techniques for antibiotic susceptibility testing. Springfield, IL: Charles C Thomas Publisher, 1973:77–87.

 36.  Tilton RC, Lieberman L, Gerlach EH. Microdilution antibiotic susceptibility test: examination of certain variables. Appl Microbiol 1973;26:658–665.

 37.  Neussel H. Problems of standardization of media. In: Williams JD, Geddes AM, eds. Chemotherapy. New York: Plenum Press, 1976:2.

 38.  National Committee for Clinical Laboratory Standards. Protocols for evaluating dehydrated Mueller-Hinton agar. Villanova, PA: National Committee for Clinical Laboratory Standards, 1996. NCCLS document M6-A.

 39.  Pursiano TA, Misiek M, Leitner F, et al. Effect of assay medium on the antibacterial activity of certain penicillins and cephalosporins. Antimicrob Agents Chemother 1973;3:33–39.

 40.  Drobot GR, Karlowsky JA, Hoban DJ, et al. Antibiotic activity in microbiological media versus that in human urine: comparison of ampicillin, ciprofloxacin, and trimethoprim-sulfamethoxazole. Antimicrob Agents Chemother 1996;40:237–240.

 41.  Hoeprich PD, Barry AL, Fay GD. Synthetic medium for susceptibility testing. Antimicrob Agents Chemother 1970;1969:494–497.

 42.  Sabath LD, Lorian V, Gerstein D, et al. Enhancing effect on alkalinization of the medium on the activity of erythromycin against Gram-negative bacteria. Appl Microbiol 1968;16:1288–1292.

 43.  Lorian V, Sabath LD. Effect of pH on the activity of erythromycin against 500 isolates of Gram-negative bacilli. Appl Microbiol 1970;20:754–756.

 44.  Bartlett JG. Campylobacter pylori: fact or fancy. Gastroenterology 1988;94:229–232.

 45.  McNulty CAM, Dent J, Wise R. Susceptibility of clinical isolates of Campylobacter pyloridis to 11 antimicrobial agents. Antimicrob Agents Chemother 1985;28:837–838.

 46.  Grayson ML, Eliopoulos GM, Ferraro MJ, et al. Effect of varying pH on the susceptibility of Campylobacter pylori to antimicrobial agents. Eur J Clin Microbiol Infect Dis 1989;8:888–889.

 47.  Debets-Ossenkopp YJ, Namarar F, MacLaren DM. Effect of an acidic environment on the susceptibility of Helicobacter pylori to trospectomycin and other antimicrobial agents. Eur J Clin Microbiol Infect Dis 1995;14:353–355.

 48.  Mackowiak RA, Marling-Cason M, Cohen RL. Effects of temperature on antimicrobial susceptibility of bacteria. J Infect Dis 1982;145:550–555.

 49.  Lamb JW, Mann JM, Simmons RJ. Factors influencing the microbiological assay of tobramycin. Antimicrob Agents Chemother 1972;1:323–328.

 50.  Wick WE. Delineation of the differences of various bacterial susceptibility tests with cephalexin. Antimicrob Agents Chemother 1969;1968:435–441.

 51.  Selwyn S, Lam C. Combined effect on protein-binding and β-lactamases on the activity of penicillins and cephalosporins. In: Williams JD, Gedes AM, eds. Chemotherapy. Vol 2. New York: Plenum Press, 1976:325–330.

 52.  Wick WE. Influence of antibiotic stability on the results of in vitro testing procedures. J Bacteriol 1964;87:1162–1170.

 53.  Sherris JC, Rashad AL, Lighthart GA. Laboratory determination of antibiotic susceptibility to ampicillin and cephalothin. Ann NY Acad Sci 1967;145:248–267.

 54.  Craig WA, Gudmundsson S. The postantibiotic effect. In: Lorian V, ed. Antibiotics in laboratory medicine. 2nd ed. Baltimore: Williams and Wilkins, 1985:515–536.

 55.  Eagle H, Musselman A. The rate of bactericidal action of penicillin in vitro as a function of its concentration, and its paradoxically reduced activity at high concentrations against certain organisms. J Exp Med 1948;88:99–131.

 56.  McDonald PJ, Craig WA, Kunin CM. Brief antibiotic exposure and effect on bacterial growth. In: Williams JD, Geddes AM, eds. Chemotherapy. Vol 2. New York: Plenum Press, 1976:95–102.

 57.  Kenny MA, Pollock HM, Minshew BH, et al. Cation components of Mueller-Hinton agar affecting testing of Pseudomonas aeruginosa susceptibility to gentamicin. Antimicrob Agents Chemother 1980;17:55–62.

 58.  Garrod LP, Waterworth PM. Effect of medium composition on the apparent sensitivity of Pseudomonas aeruginosa to gentamicin. J Clin Pathol 1969;22:534–538.

 59.  Fass RJ, Barnishan J. Effect of divalent cation concentrations on the antibiotic susceptibilities of nonfermenters other than Pseudomonas aeruginosa. Antimicrob Agents Chemother 1979;16:434–438.

 60.  Reller LB, Schoenknecht FD, Kenny MA, et al. Antibiotic susceptibility testing of Pseudomonas aeruginosa: selection of a control strain and criteria for magnesium and calcium content in media. J Infect Dis 1974;130:454–463.

 62.  Weinberg ED. The mutual effects of antimicrobial compounds and metallic cations. Bacteriol Rev 1957;21:46–68.

 63.  Medeiros AA, O’Brien T, Wacker WEC, et al. Effect of salt concentration on the apparent in vitro susceptibility of Pseudomonas and other Gram-negative bacilli to gentamicin. J Infect Dis 1971;124(Suppl):S59–S64.

 64.  Neu H. Mecillinam, a novel penicillanic acid derivative with unusual activity against Gram-negative bacteria. Antimicrob Agents Chemother 1976;9:973–999.

 65.  Jorgensen JH, Redding JS, Maher LA, et al. Improved medium for antimicrobial susceptibility testing of Haemophilus influenzae. J Clin Microbiol 1987;25:2105–2113.

 66.  Eliopoulos GM, Reiszner E, Willey S, et al. Effect of blood product medium supplements on the activity of cefotaxime and other cephalosporins against Enterococcus faecalis. Diagn Microbiol Infect Dis 1989;12:149–156.

 67.  Coudron PE, Stratton CW. Factors affecting growth and susceptibility testing of Helicobacter pylori in liquid media. J Clin Microbiol 1995;33:1028–1030.

 68.  Neu HC. Penicillins: new insights into their mechanisms of activity and clinical use. Bull NY Acad Med 1982;58:681–695.

 69.  Van Klingeren B, Rutgers A. Usefulness of commercially available media to MIC determinations of trimethoprim. In: Williams JD, Geddes AM, eds. Chemotherapy. Vol 2. New York: Plenum Press, 1976:2.

 70.  Waterworth PM. Sensitivity tests with trimethoprim-sulphonamide. S Afr Med J 1970;44(Suppl):10–12.

 71.  Then R. Thymidine and the assessment of co-trimoxazole. In: Williams JD, Geddes AM, eds. Chemotherapy. New York: Plenum Press, 1976:2.

 72.  Ferone RS, Bushby RM, Burchall JJ, et al. Identification of Harper-Cawston factor as thymidine phosphorylase and removal from media of substances interfering with susceptibility testing to sulfonamides and diaminopyrimidines. Antimicrob Agents Chemother 1975;7:91–98.

 73.  Dougherty PF, Yotter DW, Matthews TR. Chemically defined medium for susceptibility testing of antimicrobial agents. Antimicrob Agents Chemother 1976;10:923–925.

 74.  Brook I. Inoculum effect. Rev Infect Dis 1989;11:361–368.

 75.  Corrado ML, Landesman SH, Cherubin CE. Influence of inoculum size on activity of cefoperazone, cefotaxime, moxalactam, piperacillin, and N-formimidoyl thienamycin (MK 0787) against Pseudomonas aeruginosa. Antimicrob Agents Chemother 1980;18:893–896.

 76.  Seligman SJ, Hewitt WL. Resistance to penicillins and cephalosporins. Antimicrob Agents Chemother 1965;5:387–391.

 77.  Thornsberry C. Methicillin-resistant (heteroresistant) staphylococci. Antimicrob Newslett 1984;1:43–47.

 78.  Minami S, Yotsuji A, Inoue M, et al. Induction of β-lactamase by various β-lactam antibiotics in Enterobacter cloacae. Antimicrob Agents Chemother 1980;18:382–385.

 79.  Sanders CC, Sanders WE. Emergence of resistance to cefamandole: possible role of cefoxitin-inducible S. aureus lactamases. Antimicrob Agents Chemother 1979;7:1521.

 80.  Sanders CC, Sanders WE. Emergence of resistance during therapy with newer (-lactam antibiotics: role of inducible lactamases and implications for the future. Rev Infect Dis 1983;5:639–648.

 81.  Woods DD. The relation of p-aminobenzoic acid to the mechanism of action of sulphanilamide. Br J Exp Pathol 1940;21:74–90.

 82.  Chapman SW, Steigbigel RT. Staphylococcal (β-lactamase and efficacy of (β-lactam antibiotics: in vitro and in vivo evaluation. J Infect Dis 1983;143:1078–1089.

 83.  Clark H, Turck M. In vitro and in vivo evaluation of cephalexin. Antimicrob Agents Chemother 1969;8:296–301.

 84.  Kind AC, Kestle DG, Standiford HC, et al. Laboratory and clinical experience with cephalexin. Antimicrob Agents Chemother 1969;8:361–365.

 85.  Eickhoff TC, Ehret JM. In vitro comparison of cefoxitin, cefamandole, cephalexin, and cephalothin. Antimicrob Agents Chemother 1976;9:994–999.

 86.  Emerson BB, Smith AL, Harding AL, et al. Haemophilus influenzae type B susceptibility of seventeen antibiotics. Pediatrics 1975;86:617–620.

 87.  Syriopoulou V, Scheifele DW, Sack CM, et al. Effect of inoculum size on the susceptibility of Haemophilus influenzae b to β-lactam antibiotics. Antimicrob Agents Chemother 1979;16:510–513.

 88.  Barry AL, Joyce LJ, Adams AP. Rapid determination of antimicrobial susceptibility for urgent clinical situations. Am J Clin Pathol 1973;59:693–699.

 89.  Barry AL, Badal RE, Hawkinson RW. Influence of inoculum growth phase on microdilution susceptibility tests. J Clin Microbiol 1983;18:645–651.

 90.  Aldridge KE, Janney A, Sanders CV, et al. Interlaboratory variation of antibiograms of methicillin-resistant and methicillin-susceptible Staphylococcus aureus strains with conventional and commercial testing systems. J Clin Microbiol 1983;18:142–147.

 91.  Stratton CW. Serum bactericidal test. Clin Microbiol Rev 1988;1:1926.

 92.  MacLowry JD, Young MJ, Selepak ST, et al. A semiautomated microtechnique for serial dilution antimicrobial sensitivity testing in the clinical laboratory. Int J Clin Pharmacol 1970;3:70–72.

 93.  Lorian V. Antibiotics and chemotherapeutic agents in clinical and laboratory practice. Springfield, IL: Charles C Thomas, 1966.

 94.  Gerlach EH. Microdilution, I: a comparative study. In: Balows A, ed. Current techniques for antibiotic susceptibility testing. Springfield, IL: Charles C Thomas, 1973:63–76.

 95.  Gerlach EH. Dilution test procedures for susceptibility testing. In: Bondi A, Bartola JT, Prier JD, eds. The clinical laboratory as an aid in chemotherapy of infectious disease. Baltimore: University Park Press, 1977:45–50.

 96.  Gavan TL, Butler D. An automated microdilution method for antimicrobial susceptibility testing. In: Balows A, ed. Current techniques for antibiotic susceptibility testing. Springfield, IL: Charles C Thomas, 1973:88–93.

 97.  Jones RN, Gavan TL, Barry AL. Evaluation of the Sensititre microdilution antibiotic susceptibility system against recent clinical isolates: three-laboratory collaborative study. J Clin Microbiol 1980;11:426–429.

 98.  Jones RN, Thornsberry C, Barry AL, et al. Evaluation of the Sceptor microdilution investigation. J Clin Microbiol 1981;13:184–194.

 99.  McMaster PRB, Robertson E, Witebsky F, et al. Evaluation of a dispensing instrument (Dynatech MIC-2000) for preparing microtiter antibiotic plates and testing potency during storage. Antimicrob Agents Chemother 1978;13:842–844.

100.  Waterworth PM. In: Reeves DS, Phillips I, Williams JD, et al, eds. Laboratory methods in chemotherapy. Edinburgh: Churchill Livingstone, 1978:4.

101.  Woolfrey BF, Ramadel WA, QUail CO. Evaluation of a semiautomated micro-broth dilution system for determining minimum inhibitory concentrations of antimicrobics. Am J Clin Pathol 1980;73:374–379.

102.  Thornsberry C, Gavan TL, Gerlach EH. New developments in antimicrobial agent susceptibility testing. Cumitech 6. Washington, DC: American Society for Microbiology, 1977.

103.  Gavan TL. Broth dilution methods. In: Gavan TL, Cheatle EL, McFadden HW, eds. Antimicrobial susceptibility testing. Chicago: American Society of Clinical Pathology, 1971:105–124.

104.  Jackson GG, Finland M. Comparison of methods for determining sensitivity of bacteria to antibiotics in vitro. Arch Intern Med 1951;88:446–460.

105.  Waisbren BA, Carr C, Dunnette J. The tube dilution method of determining bacterial sensitivity to antibiotics. Am J Clin Pathol 1951;21:884–891.

106.  Washington JA II. Laboratory procedures in clinical microbiology. Boston: Little, Brown and Co, 1974.

107.  Jones RN, Barry AL, Gavan TL, et al. Susceptibility tests: microdilution and macrodilution broth procedures. In: Lennette EH, Balows A, Hausler WJ Jr, et al, eds. Manual of clinical microbiology. 4th ed. Washington, DC: American Society for Microbiology, 1985.

108.  Miller JM, Andersen BV, Caudill SP. New quality control frequency guidelines for antimicrobic susceptibility testing. Antimicrob Newslett 1986;3:77–79.

109.  Lebek G, Zund P. R-factors affecting host generation time: their influence on sensitivity testing. Infection 1981;9:76.

110.  Richmond MH, Sykes RB. The β-lactamases of Gram-negative bacteria and their possible physiological role. In: Rose AH, Tempesti DW, eds. Advances in microbiology and physiology. Vol 9. New York: Academic Press, 1973:31–88.

111.  Elwell LP, DeGraff J, Seibert D, et al. Plasmid-linked ampicillin resistance in Haemophilus influenzae type b. Infect Immun 1975;12:404–410.

112.  Elwell LP, Roberts M, Mayer L, et al. Plasmid-mediated (β-lactamase production in Neisseria gonorrhoeae. Antimicrob Agents Chemother 1977;11:533–538.

113.  Zigheiboin S, Tomasz A. Penicillin binding proteins of multiply antibiotic-resistant South African strains of Streptococcus pneumoniae. Antimicrob Agents Chemother 1980;17:434–441.

114.  Clinical and Laboratory Standards Institute. Performance standards for antimicrobial susceptibility testing. Nineteenth informational supplement. Wayne, PA: Clinical and Laboratory Standards Institute, 2009. CLSI publication M100-S19.

115.  Baker CN, Thornsberry C, Jones RN. In vitro antimicrobial activity of cefoperazone, cefotaxime, moxalactam (LY 127935), azlocillin, mezlocillin, and other (β-lactam antibiotics against Neisseria gonorrhoeae and Haemophilus influenzae, including (β-lactamase–producing strains. Antimicrob Agents Chemother 1980;17:757–761.

116.  Needham CA. Haemophilus influenzae: antibiotic susceptibility. Clin Microbiol Rev 1988;1:218–227.

117.  Thornsberry C, Baker CN, Kirven LA, et al. Susceptibility of ampicillin-resistant Haemophilus influenzae to seven ampicillins. Antimicrob Agents Chemother 1976;9:70–73.

118.  Doern GV, Jorgensen JH, Thornsberry C, et al. National collaborative study of the prevalence of antimicrobial resistance among clinical isolates of Haemophilus influenzae. Antimicrob Agents Chemother 1988;32:180–185.

119.  Azemun P, Stall T, Roberts M, et al. Rapid detection of chloramphenicol resistance in Haemophilus influenzae. Antimicrob Agents Chemother 1981;20:168–170.

120.  Jorgensen JH, Doern GV, Thornsberry C, et al. Susceptibility of mutliple-resistant Haemophilus influenzae to newer antimicrobial agents. Diagn Microbiol Infect Dis 1988;9:27–32.

121.  Dillon LK, Howe SE. Early detection of oxacillin-resistant staphylococcal strains with hypertonic broth diluent for microdilution panels. J Clin Microbiol 1984;19:473–476.

122.  McDougal LK, Thornsberry C. The role of (β-lactamase in staphylococcal resistance to penicillinase-resistant penicillins and cephalosporins. J Clin Microbiol 1986;23:832–839.

123.  Thornsberry C, McDougal LK. Successful use of broth microdilution in susceptibility tests for methicillin-resistant (heteroresistant) staphylococci. J Clin Microbiol 1983;18:1084–1091.

124.  Jones RN, Gavan TL, Thornsberry C, et al. Standardization of disk diffusion and agar dilution susceptibility tests for Neiserria gonorrhoeae: interpretive criteria and quality control guidelines for ceftriaxone, penicillin, spectinomycin, and tetracycline. J Clin Microbiol1989;27:2758–2766.

125.  Tenover FC, Baker CN, Swenson JM. Evaluation of commerical methods for determining antimicrobial susceptibility of Streptococcus pneumoniae. J Clin Microbiol 1996;34:10–14.

126.  Cooksey RC, Swenson JM. In vitro antimicrobial inhibition patterns of nutritionally variant streptococci. Antimicrob Agents Chemother 1979;16:514–518.

127.  Winslow DL, Damme J, Dieckman E. Delayed bactericidal activity of β-lactam antibiotics against Listeria monocytogenes: antagonism of chloramphenicol and rifampin. Antimicrob Agents Chemother 1983;23:555–558.

128.  Stevens DL, Higbee JW, Oberhofer TR, et al. Antibiotic susceptibilities of human isolates of Pasteurella multocida. Antimicrob Agents Chemother 1979;16:322–324.

129.  Vanhoof R, Gordts B, Diedickx R, et al. Bacteriostatic and bactericidal activities of 24 antimicrobial agents against Campylobacter fetus subsp jejuni. Antimicrob Agents Chemother 1980;18:118–121.

130.  Goldstein EJC, Cherubin CE, Shulman M. Comparison of microtiter broth dilution and agar dilution methods for susceptibility testing of Eikenella corrodens. Antimicrob Agents Chemother 1983;23:42–45.

131.  Jao RL, Finland M. Susceptibility of Mycoplasma pneumoniae to 21 antibiotics in vitro. Am J Med Sci 1967;253:639–650.

132.  Smith TF. In vitro susceptibility of Ureaplasma urealyticum to rosaramicin. Antimicrob Agents Chemother 1979;16:106–108.

133.  Spaepern MS, Kundsin RB. Simple direct broth-disk method for antibiotic susceptibility testing of Ureaplasma urealyticum. Antimicrob Agents Chemother 1977;11:267–270.

134.  Bowie WR. Lack of in vitro activity of cefoxitin, cefamandole, cefuroxime, and piperacillin against Chlamydia trachomatis. Antimicrob Agents Chemother 1982;21:339–340.

135.  Hammerschlag M, Gleyzer A. In vitro activity of a group of broad-spectrum cephalosporins and other (β-lactam antibiotics against Chlamydia trachomatis. Antimicrob Agents Chemother 1983;23:493–494.

136.  Brown-Elliott BA, Brown JM, Conville PS, et al. Clinical and laboratory features of the Nocardia spp. Based on current molecular taxonomy. Clin Microbiol Rev 2006;19:259–282.

137.  Verroken A, Janssens M, Berhin C, et al. Evaluation of matrix-assisted laser description ionization-time of flight mass spectrometry for identification of Nocardia species. J Clin Microbiol 2010;48:4015–4021.

138.  Clinical and Laboratory Standards Institute. Susceptibility testing of mycobacteria, nocardiae, and other aerobic actinomycetes; approved standard—2nd ed. Wayne, PA: Clinical and Laboratory Standards Institute. CLSI publication M24-A2.

139.  Malhotra-Kumar S, Haccuria K, Michiels M, et al. Current trends in rapid diagnostics for methicillin-resistant Staphylococcus aureus and glycopeptide-resistant Enterococcus species. J Clin Microbiol 2008;46:1577–1587.

140.  Rong SL, Leonard SN. Heterogenous vancomycin resistance in Staphylococcus aureus: a review of epidemiology, diagnosis and clinical significance. Ann Pharmacol 2010;44:844–850.

141.  Hiramatsu K, Hanaki H, Ino T, et al. Methicillin-resistant Staphylococcus aureus clinical strain with reduced vancomycin susceptibility. J Antimicrob Chemother. 1997;40:135–136.

142.  Swenson JM, Ferraro MJ, Sahm DF, et al. National Committee for Clinical and Laboratory Standards Study Group on Enterococci. Multi-laboratory evaluation of screening methods for detection of high-level aminoglycoside resistance in enterococci. J Clin Microbiol1995;33:3008–3018.

143.  Clinical and Laboratory Standards Institute. Performance standards for antimicrobial susceptibility testing. 21st informational supplement. Wayne, PA: Clinical and Laboratory Standards Institute, 2012. CLSI publication M100-S22.

144.  Clinical and Laboratory Standards Institute. Performance standards for antimicrobial susceptibility testing. Twenty-third informational supplement. Wayne, PA: Clinical and Laboratory Standards Institute, 2013. CLSI publication M100-S23.

145.  European Committee on Antimicrobial Susceptibility Testing. Expert rules. Version 2.0. www.eucast.org/expertrules. Accessed July 15, 2013.

146.  Queenan AM, Bush K. Carbapenemases: the versatile β-lactamases. Clin Microbiol Rev 2007;20:440–458.

147.  Villegas MV, Lolans K, Correa A, et al. First identification of Pseudomonas aeruginosa isolates producing a KPC-type carbapenem-hydrolyzing β-lactamase. Antimicrob Agents Chemother 2007;51:1553–1555.

148.  Clinical and Laboratory Standards Institute. Performance standards for antimicrobial susceptibility testing. Nineteenth informational supplement. Wayne, PA: Clinical and Laboratory Standards Institute, 2009. CLSI publication M100-S19.

149.  Livermore DM, Andrews JM, Hawkey PM, et al. Are susceptibility tests enough, or should laboratories still seek ESBLs and carbapenemeases directly? J Antimicrob Chemother 2012;67:1569–1577.

150.  Duguid JP. The sensitivity of bacteria to the action of penicillin. Edinb Med J 1946;53:401–412.

151.  Gardner AD. Morphological effects of penicillin on bacteria. Nature 1940;146:837–838.

152.  Greenwood D, Eley A. A turbidimetric study of the responses of selected strains of Pseudomonas aeruginosa to eight antipseudomonal β-lactam antibiotics. J Infect Dis 1982;145:110–117.

153.  Lorian V. Some effects on subinhibitory concentrations of antibiotics on bacteria. Bull NY Acad Med 1975;51:1046–1055.

154.  Lorian V. Effect of low antibiotic concentrations on bacteria. In: Lorian V, ed. Antibiotics in laboratory medicine. 2nd ed. Baltimore: Williams and Wilkins, 1986:596–668.

155.  Zanon U. Sub-inhibitory levels of antibiotics. J Antimicrob Chemother 1977;3:106–107.

156.  Lorian V, Waluschka A, Carruth C, et al. A five-hour disc susceptibility test. In: Lorian V, ed. The significance of medical microbiology in the care of patients. Baltimore: Williams & Wilkins, 1982.

157.  Lampe MR, Aitken CL, Dennis PG, et al. Relationship of early readings of minimal inhibitory concentrations to the results of overnight tests. Antimicrob Agents Chemother 1975;8:429–433.

158.  Sorensen RH. Rapid antibiotic sensitivity test using a redox indicator. Med Technol Bull 1959;10: 144–148.

159.  Bartlett RC, Mazens MF. Rapid antimicrobial susceptibility test using tetrazolium reduction. Antimicrob Agents Chemother 1979;15:769–774.

160.  Baker CN, Banerjee SH, Tenover FC. Evaluation of Alamar colorimetric MIC method for antimicrobial susceptibility testing Gram-negative bacteria. J Clin Microbiol 1994;32:1261–1267.

161.  Sellers W. Antibiotic sensitivity testing. In: Graber CD, ed. Rapid diagnostic methods in medical microbiology. Baltimore: Williams and Wilkins, 1970:163–178.

162.  Mitruka BM. Gas chromatographic applications in microbiology and medicine. New York: John Wiley and Sons, 1975.

163.  DeLand FH, Wagner HN Jr. Early detection of bacterial growth with carbon 14-labeled glucose. Radiology 1969;92:154–155.

164.  DeLand F, Wagner HN Jr. Automated radiometric detection of bacterial growth in blood culture. J Lab Clin Med 1970;75:529–534.

165.  Levin GV. Rapid microbiological determinations with radioisotopes. Adv Appl Microbiol 1963;5:95–133.

166.  Levin GV, Hein AH, Clendenning JR, et al. Gulliver: a quest for life on Mars. Science 1962;138:114–121.

167.  Chappelle EW, Levin GV. Use of firefly bioluminescent reaction for rapid detection and counting of bacteria. Biochem Med 1968;2:41–52.

168.  Cady P. Progress in impedance measurements in microbiology. In: Sharpe AN, Clarke DS, eds. Mechanizing microbiology. Springfield, IL: Charles C Thomas, 1978:199–239.

169.  Cady PS, Dufour W, Draeger SJ. Electrical impedance measurements: rapid method for detecting and monitoring microorganisms. J Clin Microbiol 1978;7:265–272.

170.  Ur A, Brown D. Monitoring of bacterial activity by impedance measurements. In: Heden C, Illeni T, eds. New approaches to the identification of microorganisms. New York: John Wiley and Sons, 1975:61–71.

171.  Colvin HJ, Sherris JC. Electrical impedance measurements in the reading and monitoring of broth dilution susceptibility tests. Antimicrob Agents Chemother 1977;2:61–66.

172.  Howard MC, Miyajima A, Coffman R. T-cell derived cytokines and their receptors. In: Paul WE, ed. Fundamental immunology. 3rd ed. New York: Raven Press, 1993:763–800.

173.  Johnston Laboratories. Drug susceptibility of M. tuberculosis cultures by the Bactec system: Bactec data JL1—664. In: Bactec news. Cockeysville, MD: Johnston Laboratories, 1979.

174.  McFarland J. The nephelometer: an instrument for estimating the numbers of bacteria in suspensions used for calculating the opsonic index and for vaccines. JAMA 1907;49:1176–1178.

175.  Longsworth LG. The estimation of bacterial populations with the aid of a photoelectric densitometer. J Bacteriol 1936;32:307–328.

176.  WyattPJ. Identification of bacteria by differential light scattering. Nature 1969;221:1257–1258.

177.  Wyatt PJ. Automation of differential light scattering for antibiotic susceptibility testing. In: Heden CG, Illeni T, eds. Automation in microbiology and immunology. New York: John Wiley and Sons, 1975:267–291.

178.  Amaral L, Trigenis B, Atkinson BA. The radioactive thymidine incorporation method for the determination of antibiotic susceptibility of Gram-negative bacilli. Eur J Clin Microbiol 1982;1:149–154.

179.  Pore RS. Antibiotic susceptibility testing by flow cytometry. J Antimicrob Chemother 1994;34:613–627.

180.  National Center for Infectious Diseases. Flow cytometric immunophenotyping procedure manual. Atlanta: US Department of Health and Human Services, 1993.

181.  Shapiro HM. Practical flow cytometry. 2nd ed. New York: Alan R. Liss, 1988.

182.  Nicolle Norden MA, Kurzyski TA, Bounds SE, et al. Rapid susceptibility testing of Mycobacterium tuberculosis (H37Ra) by flow cytometry. J Clin Microbiol 1995;33:1231–1237.

183.  College of American Pathologists. Proficiency survey program: College of American Pathologists surveys final critique for set D-A 199. Northfield, IL: College of American Pathologists, 1991.

184.  Amsterdam D. Instrumentation for antimicrobic susceptibility testing: yesterday, today, and tomorrow. Diagn Microbiol Infect Dis 1988;9:167–178.

185.  Staneck JL, Allen SD, Harris EE, et al. Automated reading of MIC microdilution trays containing fluorogenic enzyme substrates with the Sensititre autoreader. J Clin Microbiol 1985;22:187–191.

186.  Eng RHK, Cherubin CM, Smith SM, et al. Inoculum effect of β-lactam antibiotics on Enterobacteriaceae. Antimicrob Agents Chemother 1985;28:601–606.

187.  Elder LB, Hansen SA, Kellogg JA, et al. Verification and validation of procedures in the clinical microbiology laboratory Cumitech 31. Washington, DC: ASM Press, 1997.

188.  Doern GV, Scott DR, Rashad AL. Clinical impact of rapid antimicrobial susceptibility testing of blood culture isolates. Antimicrob Agents Chemother 1982;21:1023–1024.

189.  Trenholme GM, Kaplan RL, Karakusis PH, et al. Clinical impact of rapid identification and susceptibility testing of bacterial blood culture isolates. J Clin Microbiol 1989;27:1342–1345.

190.  Doern GV, Vaator R, Gaudet M, et al. Clinical impact of rapid in vitro susceptibility testing and bacterial identification. J Clin Microbiol 1994;32:1757–1762.

191.  Barenfanger J, Drake C, Kacich G. Clinical and financial benefits of rapid identification and antimicrobial susceptibility testing. J Clin Microbiol 1999;37:1415–1418.

192.  Schifman RB, Pindur A, Bryan JA. Laboratory practices for reporting bacterial susceptibility tests that affect antibiotic therapy. Arch Pathol Lab Med 1997;121:1168–1170.

193.  de la Maza K, Miller ST, Ferraro MJ. Use of commercially available rapid chloramphenicol acetyltransferase test to detect resistance in Salmonella species. J Clin Microbiol 1990;28:1867–1869.

194.  Tenover FC. Studies of antimicrobial resistance genes using DNA probes. Antimicrob Agents Chemother 1986;29:721–725.

195.  Rasheed JK, Tenover FC. Detection and characterization of antimicrobial resistance genes in bacteria. In: Murray PR, Baron EJ, Pfaller MA, et al, eds. Manual of clinical microbiology. 8th ed. Washington, DC: ASM Press, 2003:1196–1212.

196.  Tenover FC, Popovic T, Olsvik O. Genetic methods for detecting antibacterial resistance genes. In: Manual of clinical microbiology. 6th ed. Washington, DC: ASM Press, 1995;1368–1378.

197.  York MK, Gibbs L, Chehab F, et al. Comparison of PCR detection of mecA with standard susceptibility testing methods to detennine methicillin resistance in coagulasenegative staphylococci. J Clin Microbiol 1996;34:249–253.

198.  Dall LR, Herndon BL. Association of cell-adherent glycocalyx and endocarditis production by viridans group streptococci. J Clin Microbiol 1990;28:1698–1700.

199.  National Committee for Clinical Laboratory Standards. Proposed guidelines for pharmaceutical and susceptibility test reagent manufacturers. Villanova, PA: National Committee for Clinical Laboratory Standards, 1986. NCCLS document M23-P.

200.  Ambrose PG, Grasela DM. The use of Monte Carlo simulation to examine pharmacodynamic variance of drugs: fluoroquinolone pharmacodynamics against Streptococcus pneumoniae. Diagn Microbiol Infect Dis 2000;38:151–157.

201.  Drusano GL, Johnson DE, Rosen M, et al. Pharmacodynamics of a fluoroquinolone antimicrobial agent in a neutropenic rat model of Pseudomonas sepsis. Antimicrob Agents Chemother 1993;37:483–490.

202.  Drusano GL, Preston SL, Hardalo C, et al. Use of preclinical data for selection of a phase II/III dose for evernimicin and identification of a preclinical MIC breakpoint. Antimicrob Agents Chemother 2001;45:13–22.

203.  Mouton JW. Breakpoints: current practice and future perspectives. Int J Antimicrobial Agents 2002;19:323–331.

204.  Scavizzi MR, Bronner FD. A statistical model for the interpretation of antibiotic susceptibility tests. Int J Exp Clin Chemother 1988;1:23–42.

205.  Duan K, Dammel C, Stein J, et al. Modulation of Pseudomonas aeruginosa gene expression by host microflora through interspecies communication. Molec Microbiol 2003;50:1477–1491.

206.  Sakoulas G, Eliopoulos GM, Moellering RC Jr, et al. Staphylococcus aureus accessory gene regulator (agr) group II: is there a relationship to the development of intermediate-level glycopeptide resistance? J Clin Infect Dis 2003;187:929–938.

207.  Pillai SK, Sakoulas G, Gold HS, et al. Prevalence of the fsr locus in Enterococcus faecalis infections. J Clin Microbiol 2002;40:2651–2652.

208.  Qin X, Singh KV, Weinstock GM, et al. Effects of Enterococcus faecalis fsr genes on production of gelatinase and a serine protease and virulence. Infect Immun 200;68:2579–2586.

209.  Lorian V, Burns L. Predictive value of susceptibility tests for the outcome of antibacterial therapy. J Antimicrob Chemother 1990;25:175–181.

210.  MacArthur RD, Miller M, Albertson T, et al. Adequacy of early antimicrobic antibiotic treatment and survival in severe sepsis: experience from the MONARCS trial. 2004;38:284–288.

211.  Yu VL, Chiou CC, Feldman C, et al. An international prospective study of pneumococcal bacteremia: correlation with in vitro resistance, antibiotics administered and clinical outcome. Clin Infect Dis 2003;37:230–237.

212.  Flaherty KR, Saint S, Fendrick AM, et al. The spectrum of acute bronchitis. Postgrad Med 2001;109:39–47.

213.  Lynch JP III, Fernandez JM. Clinical relevance of macrolide-resistant Streptococcus pneumoniae for community-acquired pneumonia. Clin Infect Dis 2002;34(Suppl 1):S27–S46.

214.  Alvarez-Lerma F. Modification of empiric antibiotic treatment in patients with pneumonia acquired in the intensive care unit. ICU-Acquired Pneumonia Study Group. Intensive Care Med 1996;22:387–394.

215.  Kollef MH, Sherman G, Ward S, et al. Inadequate antimicrobial treatment of infections: a risk factor for hospital mortality among critically ill patients. Chest 1999;115:462–474.

216.  Luna CM, Vujacich P, Niederman MS, et al. Impact of BAL data on the therapy and outcome of ventilator-associated pneumonia. Chest 1997;111:676–685.

217.  Rex JH, Pfaller MA. Has antifungal susceptibility come of age? Clin Infect Dis 2002;35:982–989.

218.  Tomasz A, Albino A, Zanati E. Multiple antibiotic resistance in a bacterium with suppressed autolytic system. Nature 1970;227:138–140.

219.  Hobby GL, Meyer K, Chaffee E. Observations on the mechanism of action of penicillin. Proc Soc Exp Biol Med 1942;50:281–285.

220.  Handwerger S, Tomasz A. Antibiotic tolerance among clinical isolates of bacteria. Rev Infect Dis 1985;7:368–386.

221.  Sherris JC. Problems in in vitro determination of antibiotic tolerance in clinical isolates. Antimicrob Agents Chemother 1986;30:633–637.

222.  Tuomanen E. Phenotypic tolerance: the search for β-lactam antibiotics that kill nongrowing bacteria. Rev Infect Dis 1986;8(Suppl):S279–S291.

223.  Tuomanen E, Durack DT, Tomasz A. Antibiotic tolerance among clinical isolates of bacteria. Antimicrob Agents Chemother 1986;30:521–527.

224.  Sabath LD, Wheeler N, Laverdiere M, et al. A new type of penicillin resistance of Staphylococcus aureus. Lancet 1977;1:443–447.

225.  Tipper DJ. Mode of action of beta-lactam antibiotics. In: Queener SF, Webber JA, Queener SW, eds. Beta-lactam antibiotics for clinical use. New York: Marcel Dekker Inc, 1986:17–47.

226.  Tomasz A. The mechanism of the irreversible antimicrobial effects of penicillins: how the β-lactam antibiotics kill and lyse bacteria. Annu Rev Microbiol 1979;33:113–137.

227.  Yocum RR, Rasmussin JR, Strominger SL. The mechanism of action of penicillin: penicillin activates the active site of Bacillus stearothermophilus D-alanine carboxypeptidase. J Biol Chem 1980;255:3977–3986.

228.  Blumberg PM, Strominger JL. Interaction of penicillin with the bacterial cell: penicillin-binding proteins and penicillin-sensitive enzymes. Bacteriol Rev 1974;38:291–335.

229.  Spratt BG. Biochemical and genetical approaches to the mechanism of action of penicillin. Philos Trans R Soc Lond [Biol] 1980;289:273–283.

230.  Tipper DJ, Wright A. The structure and biosynthesis of bacterial cell walls. In: Sokatch JR, Ornston LN, eds. Mechanisms of adaptations. New York: Academic Press,

231.  Georgopapadakou NH, Liu FY. Binding of β-lactam antibiotics to penicillin-binding proteins of Staphylococcus aureus and Streptococcus faecalis: relation to antibacterial activity. Antimicrob Agents Chemother 1980;18:834–836.

232.  Neu HC. Penicillin-binding proteins and role of amdinocillin in causing bacterial cell death. Am J Med 1983;75(Suppl 2A):9–20.

233.  Spratt BG. Properties of the penicillin-binding proteins of Escherichia coli K 12. Eur J Biochem 1977;72:341–352.

234.  Tomasz A. Penicillin-binding proteins and the antibacterial effectiveness of β-lactam antibiotics. Rev Infect Dis 1986;8(Suppl 3):S260–S278.

235.  Wilson WR, Geraci JE. Antibiotic treatment of infective endocarditis. Annu Rev Med 1983;34:413–427.

236.  Dormer AE. The treatment of bacterial endocarditis. Br Med Bull 1960;16:61–66.

237.  Stratton CW. The usefulness of the serum bactericidal test in orthopedic infections. Orthopedics 1984;7:1579–1580.

238.  Sculier JP, Klastersky J. Significance of serum bactericidal activity in gram-negative bacillary bacteremia in patients with and without granulocytopenia. Am J Med 1984;76:429–435.

239.  Reimer LG, Stratton C, Reller LB. Minimum inhibitory and bactericidal concentrations of 44 antimicrobial agents against three standard control strains in broth with and without human serum. Antimicrob Agents Chemother 1981;19:1050–1055.

240.  Meylan PR, Francioli P, Glauser MP. Discrepancies between MBC and actual killing of viridans group streptococci by cell-wall-active antibiotics. Antimicrob Agents Chemother 1986;29:418–423.

241.  Schoenknecht FD, Sabath LD, Thornsberry C. Susceptibility tests: special tests. In: Lennette EH, Balows A, Hausler WH Jr, et al, eds. Manual of clinical microbiology. 3rd ed. Washington, DC: American Society for Microbiology, 1985:1000–1008.

242.  National Committee for Clinical Laboratory Standards. Methods for determining bactericidal activity of antimicrobial agents. Wayne, PA: National Committee for Clinical Laboratory Standards, 1999. NCCLS document M26-A.

243.  Anhalt JP, Washington JW II. Bactericidal tests. In: Washington JA, ed. Laboratory procedures in clinical microbiology. 2nd ed. New York: Springer-Verlag, 1985:431–745.

244.  Taylor PC, Schoenknecht FD, Sherris JC, et al. Determination of minimum bactericidal concentration of oxacillin for Staphylococcus aureus: influence and significance of technical factors. Antimicrob Agents Chemother 1983;23:142–150.

245.  James PA. Comparison of four methods for the determination of MIC and MBC of penicillin for viridans streptococci and the implications for penicillin tolerance. J Antimicrob Chemother 1990;25:209–216.

246.  Hacek DM, Dressel DC, Peterson LR. Highly reproducible bactericidal activity test results by using a modified rational committee for clinical laboratory standards broth microdilution technique. J Clin Microbiol 1999;37:1881–1884.

247.  Schlichter JG, MacLean H. A method of determining the effective therapeutic level in the treatment of subacute bacterial endocariditis with penicillin. Am Heart J 1947;34:209–211.

248.  Stratton CW, Weinstein MP, Reller LB. Correlation of serum bactericidal activity with antimicrobial agent level and minimal bactericidal concentration. J Infect Dis 1982;145:160–168.

249.  National Committee for Clinical Laboratory Standards. Methodology for the serum bactericidal test. Wayne, PA: National Committee for Clinical Laboratory Standards, 1999. NCCLS document M21-A.

250.  Prober CG, Yeager AS. Use of the serum bactericidal titer to assess the adequacy of oral antibiotic therapy in the treatment of acute hematogenous osteomyelitis. J Pediatr 1979;95:131–135.

251.  Sande MA. Antibiotic therapy of bacterial meningitis: lessons we’ve learned. Am J Med 1981;7:507–510.

252.  Barriere SL, Ely E, Kapusnik JE, et al. Analysis of a new method for assessing activity of combinations of antimicrobials: area under the bactericidal curve. J Antimicrob Chemother 1985;16:49–59.

253.  Blondeau JM, Zhao X, Hansen G, et al. Mutant prevention concentrations of fluoroquinolones for clinical isolates of Streptococcus pneumoniae. Antimicrob Agents Chemother 2001;45:433–438.

254.  Dong Y, Zhao X, Domagala J, et al. Effect of fluoroquinolone concentration on selection of resistant mutants of Mycobacterium bovis BCG and Staphylococcus aureus. Antimicrob Agents Chemother 2000;

255.  Small PM, Chambers HF. Vancomycin for Staphylococcus aureus endocarditis in intravenous drug users. Antimicrob Agents Chemother 1990;34:1227–1231.

256.  Baron S, Tyring SK, Fleischman R, et al. The interferons: mechanisms of action and clinical applications. JAMA 1991;266:1375–1383.

257.  Mohr JF, Wanger A, Rex JH. Pharmacokinetic/pharmacodynamic modeling can help guide targeted antimicrobial therapy for nosocomial Gram-negative infections in critically ill patients. Diag Microbiol Infect Dis 2004;48:125–130.

258.  National Committee for Clinical Laboratory Standards. Evaluation of lots of Mueller-Hinton broth for antimicrobial susceptibility testing: proposed guideline. Wayne, PA: National Committee for Clinical Laboratory Standards, 2001. NCCLS document M32-P.

259.  Thornsberry C, Swensen JM, Baker CN, et al. Susceptibility testing of fastidious and unusual pathogens. Antimicrob Newslett 1987;4:47–55.

260.  Stull VR. Clinical laboratory use of differential light scattering, I: antibiotic susceptibility testing. Clin Chem 1973;19:883–890.

261.  Jorgensen JH, Lee JC, Jones PM. Chemically defined antimicrobial susceptibility test medium for Pseudomonas aeruginosa. Antimicrob Agents Chemother 1977;11:415–419.

262.  Baker CN, Hollis G, Thornsberry C. Antimicrobial susceptibility testing of Francisella tularensis with a modified Mueller-Hinton broth. J Clin Microbiol 1988;22:212–215.

263.  Baker CN, Thornsberry C, Hawkinson RW. Inoculum standardization in antimicrobial susceptibility testing: evaluation of overnight agar cultures and the Rapid Inoculum Standardization System. J Clin Microbiol 1983;17:450–457.

266.  Dugan DL, Wright DN. Quantitative infrared photoanalysis of selected bacteria. Appl Microbiol 1974;28:205–211.

267.  Forrest A, Nix DE, Ballow CH, et al. Pharmacodynamics of intravenous ciprofloxacin in seriously ill patients. Antimicrob Agents Chemother 1993;37:1073–1081.

269.  Inderlied CB, Young LS. Radiometric in vitro susceptibility testing of Mycobacterium tuberculosis. Antimicrob Newslett 1986;3:55–60.

270.  Wicks JH, Nelson RL, Krejcarek GE. Rapid inoculum standardization system: a novel device for standardization of inocula in antimicrobial susceptibility testing. J Clin Microbiol 1983;17:1114–1119.

272.  Kushner I. Regulation of the acute phase response by cytokines. In: Oppenheim JJ, Rossio JL, Geaniny AH, eds. Clinical application of cytokines. New York: Oxford University Press, 1993:27–34.

273.  Luria SE. A test for penicillin sensitivity and resistance in Staphylococcus. Proc Soc Exp Biol Med 1946;61:46–51.

274.  Mardh PA, Anderson KE, Wadso I. Kinetics of the actions of tetracyclines on Escherichia coli as studied by microcalorimetry. Antimicrob Agents Chemother 1976;10:604–609.

275.  Preston SL, Drusano GL, Berman AL, et al. Pharmacodynamics of levofloxacin: a new paradigm for early clinical trials. JAMA 1998;279:125–129.

276.  Pulliam L, Dall L, Inokuchi S, et al. Effects of exopolysaccharide production by viridans streptococci on penicillin therapy of experimental endocarditis. J Infect Dis 1985;151:153–156.

277.  Spratt BG. Distinct penicillin binding proteins involved in the division, elongation, and shape of Escherichia coli, K 12. Proc Natl Acad Sci USA 1975;72:2999–3003.

279.  Waxman DJ, Strominger JL. Penicillin-binding proteins and the mechanisms of action of β-lactam antibiotics. Annu Rev Biochem 1983;52:825–869.