Dawn M. Boothe, DVM, PhD, DACVIM, DACVCP
Department of Anatomy, Physiology and Pharmacology, College of Veterinary Medicine, Auburn University
Once the need for an antimicrobial is determined, the selection of the most appropriate antimicrobial begins with identifying the target organism. Selection should strive for the narrowest spectrum. For complicated infections, identification most appropriately should be based on C&S in order to identify the target, to detect resistance and to design the dosing regimen for the patient. Although culture and susceptibility data (C&S) can be a powerful tool to guide selection, it remains an in vitro test that is applied to in vivo conditions and over-reliance on the information it provides can lead to therapeutic failure. Basing antimicrobial selection on C&S data does not guarantee success, just as failing to use C&S as a basis for selection (or selecting a drug characterized by "R" on the data) does not guarantee failure. The 90-60 rule notes that approximately 90% of infections treated based on C&S are likely to respond, where as 60% will respond even if drugs noted as resistant are selected. This manuscript will address those aspects of C&S testing that can complicate interpretation some which might enhance efficacy and others that contribute to therapeutic failure.
PITFALLS OF CULTURE AND SUSCEPTIBILITY TESTING
As early as the collection process, C&S data can be misleading. It is beyond the scope of this manuscript to delineate proper techniques of specimen culture collection, but close adherence to recommended procedures including but not limited to site selection, site preparation and sample handling are critical to proper interpretation. Anaerobic infections are particularly problematic. Obligate anaerobes are exquisitely sensitive to increased oxygen tension and will not survive if exposed to oxygen. No growth may be mistakenly interpreted as lack of anaerobic infection. Many organisms are facultative anaerobes, capable of growth in anaerobic environments. Aerobic cultures may yield their growth, but the anaerobic environment in the patient may limit response to antimicrobials (particularly aminoglycosides which are ineffective in an anaerobic environment). Mycoplasma, L-forms and other organisms are difficult to culture and will not be identified with routine procedures. Not all organisms grow sufficiently rapidly for agar gel diffusion techniques. Thus, tube dilution data seldom often is not provided for anaerobic organisms. Generally, only rapidly growing aerobic organisms are conducive to testing using these methods, and particularly tube dilution methods.
Just as absence growth does not indicate lack of infection, growth of an organism should not be interpreted as evidence of infection or identification of the infecting organism. Clearly, culture of an organism from a tissue that is normally sterile indicates infection. However, discriminating between normal and infecting flora can be difficult. Pure, vibrant (meaning special media was not needed to coax the growth of the organism) of a large number of organisms are indicators of infection. A call to the diagnostic lab might be prudent before marked financial commitment is put into treating a questionable organism.
The culture and susceptibility procedures themselves are fraught with potential errors. For practices that provide in-house susceptibility testing, care must be taken to follow guidelines established and published by the National Committee on Clinical Laboratory Standards (NCCLS) (USA) or comparable federal agency. Materials, including interpretive standards, should be validated by the appropriate agency. Minor changes in pH, temperature, humidity, etc can profoundly affect results. Personnel should be trained specifically in culture techniques and hospitals that provide this service (as do diagnostic labs) should maintain well designed and adequately collected quality control data to validate their procedures (NCCLS provides control organisms). Despite the perfectly cultured organism, C&S data is inherently deficient because the in vitro methods used do not mimic in vivo conditions. For example, in the "test tube", the organism is exposed to the same conditions, including drug concentrations throughout the incubation period, a situation which does not occur in the patient.
Clearly, in vitro methods can not take into account host factors which detract from efficacy. Among the most problematic concern is the fact that determination and interpretation of minimum inhibitory concentrations (MIC and MICBP) are based on assumed plasma drug concentrations (PDC), yet infections generally are located in extracellular fluid (ECF) and occasionally intracellular. Basing MIC interpretation on PDC can result in over or under interpretation of C&S data. For tissues which concentrate the drug (or if the drug can be applied topically), and for drugs which can be concentrated by phagocytes and thus transported to the site of infection, concentrations may markedly exceed PDC, resulting in underestimation of efficacy. In such circumstances, a drug noted as "I", or , under the appropriate circumstances (e.g., topical therapy) "R" might actually be effective. Problems contributing to therapeutic failure include immunocompromise (design a dosing regimen that will assure bactericidal concentrations of the chosen drug reach the site of infection), inflammatory response (débride or otherwise appropriate clean/drain accessible infections, select a drug that distributes into tissues well and ideally accumulates in phagocytes and increase the dose appropriately), non-fenestrated capillaries (including the prostate, eye, brain, testicles, cartilage; select a drug that distributes well into tissues and increase dose as appropriate). PDC are often several fold higher than concentrations in such organs. Extra precautions should be made when designing the dosing regimen for such infections, particularly if a water soluble drugs are to be used. For example, in humans, up to 5 X the recommended dose of beta-lactams drugs is administered when treating infections of the central nervous system. Even tissues traditionally considered "well perfused" might be of concern. For example, drugs do not penetrate bronchial secretions well, despite the fact that the lungs are well perfused. Amoxicillin is often used to treat respiratory tract infections. Yet, only 3% of the amoxicillin that is in plasma is distributed to bronchial secretions. Theoretically, one must dose amoxicillin 30X the recommended dose to achieve targeted PDC in bronchial secretions. Most water soluble drugs (beta-lactams and aminoglycosides) reach only 20 to 25% of PDC in bronchial secretions whereas over 50% of lipid soluble drugs reach bronchial secretions. Dosing adjustments also are necessary for those infections that are intracellular or complicated by host response to infection.
Pitfalls of susceptibility testing also reflect the drugs selected for testing. Because automated systems can not accommodate and laboratories can not test cost effectively all potential drugs used to treat an infection, one drug often is tested as a model for other drugs in the class (i.e., cephalothin for the first generation cephalosporins; enrofloxacin for the fluorinated quinolones). Although for some classes of drugs, cross reactivity can be similar within the class (for example, an organism that is resistant or susceptible to one veterinary fluorinated quinolone is likely to be resistant or susceptible to all), care must be taken with other classes (for example ampicillin may be less effective than amoxicillin; cefazolin may be more effective against some organisms than cephalothin). Clearly, for third generation cephalosporins, differences among drugs do not allow extrapolation of data. Because many drugs in veterinary medicine are approved for use in humans rather than animals, interpretive MIC data (MICBP) is based on human, not veterinary organisms. Not only will the organisms behave differently among species, the kinetics of the drugs also are different. The MICBP also is based on Cmax in plasma, yet many infections occur in tissues.
An additional problem with interpretation is represented by drugs with active metabolites, such as enrofloxacin or ceftiofur. MIC data is based on the parent drug and does not include efficacy represented by the active metabolites. For enrofloxacin, an additional 30 to 40% Cmax or area under the curve (AUC) is contributed by conversion to ciprofloxacin. Thus, susceptibility data underestimates efficacy of enrofloxacin.
Culture and susceptibility techniques may not accurately reflect resistance that has developed in the infecting organism to drug to which the organism is generally susceptible. Beta-lactamase resistance (especially for selected cephalosporins) often is not detected by C&S data, but in vivo, organisms are able to generate enough enzyme to render the drug ineffectual. Laboratories are currently generating methods intended to detect this level of resistance.
Relationship between MIC, Plasma and Tissue Drug Concentrations
The bridging of pharmacodynamic (susceptibility) data with pharmacokinetic data should begin with an appreciation of the relationship between the MIC established by in vitro testing and the concentrations of drug achieved in the patient at the site of infection when the drug is administered at the labeled dose.
Bactericidal versus bacteriostatic drugs. The term "bactericidal" is somewhat abused: clinicians will reach for a drug that is "cidal" rather than "static", assuming that the ability of the drug to kill rather than simply inhibit an organisms will increase the risk of therapeutic efficacy. This may be true, but only if concentrations of the drug are achieved at the site of infection are sufficient to kill the microbe. The term "bactericidal" is an in vitro definition and is based on the proximity of the minimum bactericidal concentration (MBC) of a drug to the MIC. The MBC is determined following tube dilution procedures: tubes with no observable growth are inoculated on agar gel. If no organism grows on the agar, the organisms were killed in the test tube. The tube with the lowest concentration of drug that yields no growth on the agar gel contains the MBC of drug. For drugs considered "bactericidal", the MBC is within one tube dilution of the MIC, meaning, the organisms were not simply inhibited, but rather, were killed. For "bacteriostatic" drugs, growth on the agar plate will occur for several tube dilutions above the MIC, indicating that organisms were not killed. Thus, killing concentrations are more likely to be achieved than static concentrations of a drug that is bactericidal, but only if the targeted drug concentration (i.e., MIC/MBC) is achieved. The bactericidal nature of a drug reflects its mechanism of action. Drugs which target ribosomes (e.g., tetracyclines, macrolides, lincosamides, chloramphenicol) often simply inhibit the growth of the organism, and, because a much higher drug concentration is necessary to kill the organism, in vitro, the MIC is distant from the MBC. Clinically, host defenses must eradicate the infection following treatment with these drugs unless exceptionally high concentrations (i.e., the MBC) of these drugs are achieved in tissues. An exception is made for aminoglycosides, whose ribosomal inhibition is so effective that the organism dies. Drugs which target cell walls (beta lactams including penicillins and cephalosporins; vancomycin), cell membranes (bacitracin, polymixin and colistin), and DNA (enrofloxacin, metronidazole), RNA (rifampin) are defined in vitro as bactericidal. Combinations of static drugs can often result in cidal actions. For example, sulfonamides (which target folic acid synthesis) are static, but when used in combination with diaminopyrimidines (e.g., trimethoprim), the combination is defined in vitro as cidal. Attaining bactericidal concentrations of an antimicrobial is critical for those infections for which host killing is likely to be impaired. These include but are not limited to infections in immune compromised animals (e.g., viral infections [parvovirus, panleukopenia, FIV, FeLV), patients receiving glucocorticoids), or in systems characterized by derangements in local immunity (i.e., CNS infection for which an marked inflammatory response can be life threatening; osteomyelitis; peritonitis, bacteremia/sepsis, many chronic infections).
Post-antibiotic effect. The elimination half-life of many antimicrobials ranges from 1 to 4 hours. Perhaps for convenience's sake (i.e., compliance), most dosing intervals (particularly for older drugs) range from 8 to 12 hours. Thus, for many drugs, concentrations in plasma or tissue are nondetectable at the end of the dosing interval. Yet, antimicrobial efficacy may not be impaired for some drugs. Persistence of antimicrobial effects after brief exposure to (or the lack of detectable concentrations of) an antimicrobial has been termed the post-antibiotic effect (PAE). The PAE can prolong the dosing interval and is therapeutically important for some antimicrobials against some organisms. For some drugs (e.g., fluorinated quinolones and aminoglycosides), the duration of the PAE (and thus antibiotic efficacy) is concentration dependent (often referred to as dose dependent) and is maximized by a large (8-10) PDC:MIC or inhibitory quotient (the ratio of PDC:MIC). In contrast, efficacy of beta-lactams and most bacteriostatic drugs is considered time (time of exposure) dependent. For such drugs, PDC should remain above the MIC during the majority of the dosing. Reviewing the mechanism of actions of the antimicrobials may elucidate the differences. Once the target of fluorinated quinolones (DNA gyrase) and aminoglycosides (30 and 50s subunits of ribosomes) is bound to the drug, the detrimental effects of the drug on the target are irreversible. Thus, for such drugs, assuring sufficient concentrations (molecules) of drug to "wipe out" the target is critical. Once the target is impaired, no further drug is necessary. On the other hand, the beta lactams target cell wall synthesis. As new cell wall (albeit faulty) is being built, old cell wall is broken down by the organism. The organism not only must be constantly growing, but as long as new cell wall is being synthesized, the drug must be present. Thus, beta lactams should always be present. The variability in the relationship between PDC and MIC can impact the dosing regimen for a drug as is exemplified by comparing beta-lactams (interval or time dependent) with aminoglycoside and fluorinated quinolones (dose or concentration dependent) antimicrobials. Efficacy of the concentration dependent drugs is enhanced by administering a high dose which maximizes the inhibitory quotient. In contrast, efficacy of beta-lactam or bacteriostatic antimicrobials is enhanced by using shorter dosing intervals, although a dose increase may be necessary in some instances in order to surpass the MIC. Thus, using a dose that is too low is particularly detrimental with fluorinated quinolones, whereas prolonging the dosing interval should be avoided for beta-lactams.
It is important to note that reported relationships between PDC, MIC and therapeutic efficacy (and the PAE) is based on in vitro data; additionally, the relationships vary with drugs and organisms. More recent literature has proposed the area under the inhibitory curve (AUIC) as a better predictor of antimicrobial efficacy). This variable is derived by dividing the area under the PDC versus time curve by the MIC of the infecting organism; as such, both peak concentrations and elimination half-life are relevant to drug efficacy.
Enhancing Antimicrobial Efficacy. Examining the relationship between MIC and PDC more closely emphasizes the importance of selecting an adequate dose. If one assumes the recommended dose is designed to achieve the MICBP of the drug (a reasonable assumption as the MICBP is based, in part, on the Cmax), clearly this concentration will be insufficient for concentration-dependent drugs (for which PDC should be 8-10 X the MIC) unless the MIC of the infecting organisms is approximately 1/10th of the MICBP. For example, whereas a dose of enrofloxacin at 5 mg/kg (which achieved PDC of 1 μg/ml) might be sufficient to treat an E coli with an MIC of 0.06 μg/ml (0.06 μg/ml * 8 = 0.48 μg/ml which is < 1 μg/ml), a dose of even 20 mg/kg, which achieves PDC of approximately 4 μg/ml may not be sufficient to treat a Pseudomonas aeruginosa organisms with an MIC of 1 μg/ml even though a C&S report might indicate "S" (1 μg/ml * 8 = 8 μg/ml). Time dependent drugs do not require 8-10 * the MIC for efficacy. However, they do require PDC above the MIC for most of the dosing interval. Yet, if the dose is designed to simply achieve the MICBP of the drug, for organisms whose MIC is close to the MICBP, PDC will rapidly drop below the MIC. For example, if a Staphylococcus aureus has an MIC of 8 μg/ml for a beta-lactam, with an MICBP of 32 μg/ml, only two half-lives of the drug can elapse before the PDC has dipped below the MIC (PDC = 16 μg/ml after the first half-life, and 8 μg/ml after the second). Most beta-lactams have half-lives of 2 hours or less, leaving 4 hours between dosing intervals. Doubling the dose of the drug will add one more half-life to the dosing interval. Thus, for time dependent drugs, both the dose (increased) and the interval (shortened) may need to be altered for organisms whose MIC is close to the MICBP. This approach is based on drug concentrations; modifications become more important and potentially greater for infections further complicated by distribution site, etc.
Mutant Potential Concentration (MPC): DEAD BUGS DON'T MUTATE. The use of low concentrations of an antimicrobial for a long period of time (> 10-14 days) is the ideal way to nurture the development and expansion of mutant microbes characterized by antimicrobial resistance. The MPC refers to the minimal antibiotic concentration that prevents the selection of first-step resistant mutants in the presence of large numbers of cells (1010). Determination of the MPC requires separating out from other colonies those colonies of a bacterial strain which are characterized by high MIC (i.e., exceeds the MIC 90 of the strain), a tedious and not generally available technique. Unfortunately, the MIC of a drug does not predict the MPC of the organism. However, increasingly studies are reporting the likelihood that simply achieving the MIC of an isolated bacteria may contribute to resistance whereas achieving the MPC (which will be substantially higher than the MIC) will reduce the development of resistance.
Achieving adequate drug concentrations at the site of infection. The MIC is based on plasma drug concentrations yet infection is in the tissue. Reaching adequate concentrations at the site clearly is an important goal of antimicrobial therapy. Actions that can be taken to achieve effective concentrations include modifications for both the dose (increase) of both time and concentration dependent drug and intervals (shorten) for time dependent drugs. Other modifications would include administration of the drug intravenously to assure 100% bioavailability as well as generation of rapid (and thus higher) peak plasma drug concentrations (particularly important for concentration-dependent drugs), and combination antimicrobial therapy. Efficacy will be facilitated by selection of a drug characterized by adequate distribution (lipid soluble [volume of distribution > 0.6 L/kg). For example, porin penetration can be enhanced by selection of a lipid soluble drug (most drugs EXCEPT beta-lactams [penicillins, cephalosporins] and aminoglycosides); doxycycline or minocycline rather than tetracyclines). Movement into the microbe can be facilitated by smaller molecular weight drugs (e.g., extended spectrum penicillins such as ticarcillin, imipenem). Use of a drug accumulated by phagocytes will enhance distribution to the site of infection. These include any fluorinated quinolone, clindamycin, and the macrolides (erythromycin, azithromycin).
Combination antimicrobial therapy may be the single most effective action taken to enhance antimicrobial efficacy in the chronic or serious infection. Combination therapy also may reduce the advent of resistance. The combination of two drugs noted as "I" or even "R" together can render the drug susceptible. Current C&S techniques do not consider the use of drugs in combination. Note however, that research supporting combination therapy also is in vitro and may not reflect clinical conditions. Nonetheless, clinically, combination therapy can be a powerful tool. Antimicrobials to be used in combination therapy should be selected rationally and should be based on target organisms as well as mechanism of action In general, "bacteriostatic" drugs which inhibit ribosomes and thus microbial growth (e.g., chloramphenicol, tetracyclines, and erythromycin) should not be combined with drugs whose mechanism of action is dependent upon protein synthesis, e.g., growth of the organism (e.g., beta-lactams) or formation of a target protein. The bactericidal activity of and continued degradation or destruction of the microbial target β-lactams and fluorinated quinolones depends on continued synthesis of bacterial proteins. Antagonistic effects have been well documented between β-lactam antimicrobials and inhibitors of ribosomal activity. Generally, drugs that have the same mechanism of action probably act in an additive fashion. For example, ciprofloxacin is an active metabolite generated from metabolism of enrofloxacin in most species. Together, these compounds act in an additive fashion. Synergism between antimicrobials is most likely to occur if the two antimicrobials kill bacteria through independent mechanisms or through sequential pathways towards the same target. Four well established mechanisms of synergism exist: sequential inhibition of a common pathway; inhibition of an enzyme of destruction (e.g., beta-lactamase), sequential inhibition of cell wall synthesis, or facilitation of entry of an antimicrobial through the cell wall. The advantage of a β-lactam combination may reflect increased concentrations of the second drug as a result of inhibition of cell wall synthesis by the first drug (β-lactam). Synergism between β-lactams and aminoglycosides exemplifies synergism due to killing by independent pathways. Synergism is expected because their mechanisms of action compliment one another, but efficacy is enhanced further because aminoglycoside movement into the bacteria is enhanced by increased cell wall permeability induced by the β-lactam. Enhanced movement in a bacteria may occur for other drugs (e.g., potentiated sulfonamides) when combined with β-lactams. The combination of trimethoprim and a sulfonamide exemplifies synergism resulting from sequential actions in the same metabolic pathway.