Drug Selection And Dosing

If a specific etiology is not or cannot be determined, treatment must be determined on the basis of clinical trials (proof of efficacy) and knowledge of usual etiologies. Clinical trials of anti-bacterial agents do not typically address optimal therapy for a particular infection. Given design constraints of including only patients who are likely to respond (i.e., have susceptible organism and no conditions that complicate response) and study goals of demonstrating equivalence between treatments, few treatments are demonstrated to be superior (19). Clinical trials are focused on a specific disease, such as community-acquired pneumonia or intra-abdominal infection. Consideration of etiology is usually a secondary concern unless in vitro resistance is documented at protocol entry. Antibacterial drugs are labeled for indications that include specific pathogens based on an adequate number of cases entered into phase III clinical trials. However, such label information does not ensure that the antibacterial agent is the best available drug. In many cases, the science available may support nonlabeled uses for the agent. A pharmacodynamic (PD) approach considers the etiology (pathogen susceptibility) and antibacterial exposure (pharmacokinetics [PK]) as most important. Excluding specialized infection sites such as the cerebrospinal space, middle ear space, or urine, serum concentrations may be used as a surrogate for concentrations at the site of infection. For agents that have significant protein binding (particularly agents that are 85% or more bound), the free concentration represents the bioactive concentration and should be used for defining exposure (20). Special considerations are needed for bacteria that survive and multiply within cells, such as Chlamydia spp. and Legionella spp (21). Likewise, PK/PD relationships may differ substantially among agents. Free serum concentrations of a beta-lactam agent may predict response to infections caused by extracellular pathogens, for example. However, agents that have high tissue affinity (e.g., azithromycin) may have greater efficacy than would be predicted based on serum concentrations (22). Optimally, one would have PK/PD studies performed for each type of infection and groups of antibacterial drugs and would understand the relationships among dose, biophase exposure, antibacterial activity, and clinical efficacy. Whether antibacterial activity is predictable on the basis of antibacterial exposure and pathogen susceptibility would be established. Common organisms that behave differently than predicted would also be identified. Our knowledge of PK/PD is clearly lacking for many infections; however, the information available does provide a framework for designing optimal therapy. Prevention of medication errors and adverse events involving antibacterial drugs requires a broad-based, integrated effort. Potential adverse effects would be avoided in patients who do not receive antibacterial therapy when such therapy is not indicated. These adverse effects include not only adverse effects in the individual patient, but also adverse effects to society, such as increasing bacterial resistance and increased health care costs. However, the risk of overtreatment should be balanced with the risk of undertreatment. Failure to administer an antibacterial agent to patients who have an active infection and could benefit from antibacterial therapy could lead to more serious illness and long-term and/or permanent complications. Thus, the first principle is to use antibacterial drugs in appropriate circumstances. The second principle is to administer the right drug in appropriate clinical situations. Achieving this second principle is less defined. Is the right drug one that is Food and Drug Administration (FDA) approved for the specific indication, generally accepted as effective for the specific indication, the drug of choice for the specific indication, or the best drug for the individual patient? Clinical trials, as currently performed, are not designed to provide the information needed to promote optimal use of antibacterial drugs in individual patients. Obstacles to individualized antibacterial selection and dosing include failure to identify a specific pathogen (s), uncertainty of etiology, semiquantitative susceptibility testing, lack of predictive pharmacokinetic models or analytical capabilities, and uncertainties as to whether more scientifically based antibacterial prescribing will enhance outcomes. The application of PK/PD strategies has not been demonstrated to be cost-effective. In the end, the desire to have simplified marketing and dosing is a major obstacle to using a PK/PD approach.

Pharmacodynamic principles allow for individualization of drug selection. Clinicians will need to be familiar with pharmacokinetics as they relate to drug exposure at different sites of infection, in vitro susceptibility, and pharmacodynamic goals as they relate to different classes of antibacterial drugs. Exposure parameters such as maximum drug concentration/ minimum inhibitory concentration (Cmax/MIC) ratio are useful for many drugs, including aminoglycosides and fluoroquinolones. Here, the population Cmax may be used with the MIC90 (MlC for 90%) of the species or with the individualized pathogen MIC. In the original description, the agent with the highest Cmax/MIC would be selected (23). In reality, there may be several agents that have Cmax/MIC ratios that are sufficient for optimal response, and other factors (cost, dosing convenience, safety) can be used to select between the agents. For aminoglycosides and fluoroquinolones, Cmax/MIC ratios of 8 to 10 or greater are considered optimal for most infections (24,25). Generally the serum Cmax can be used except when the site of infection involves a barrier to drug penetration. For example, cerebrospinal fluid (CSF) concentration should be used with meningeal infection and middle ear fluid concentrations for the treatment of otitis media (26). The area under the plasma concentration-time curve divided by MIC (AUC/ MIC) is also useful for optimal use of fluoroquinolones. In fact, this parameter predicts outcome slightly better than Cmax/ MIC. The goal for AUC/MIC may vary depending on the specific fluoroquinolone and specific infection; however, in general, the AUC/MIC should optimally exceed 100 to 125 for gram-negative infections (25).

Antimicrobial activity is best explained by the free drug concentration and exposure parameters based on free drug concentrations. However, many of the clinical PK/PD study reports provide conclusions based on total drug exposure rather than free (unbound) drug exposure. A better understanding of PK/PD targets is needed before extrapolating a study involving one drug to other drugs in the same class. Differences in penetration to the site of infection, protein binding, and other undetermined factors may affect the exposure parameter target values.

After the correct drug is selected, dosing needs to be optimized using the same pharmacodynamic principles. The susceptibility breakpoints for gentamicin are 4 pg/mL or greater for susceptible pathogens and 8 pg/mL for intermediate pathogens (27). Current recommendations for high-dose extended-interval dosing target maximum concentrations of 14 to 20 pg/ mL. A pathogen could be reported susceptible to gentamicin, whereas maximum doses would achieve a Cmax/MIC ratio of only 3.5 to 5 and AUC/MIC of 13 to 19. Optimal pharmacodynamics is believed to be associated with a Cmax/MIC ratio of 10 or greater and AUC/MIC of 70 or greater (24,28). Use of gentamicin in this setting would be acceptable only in combination with a primary antibacterial agent that would achieve more optimal pharmacodynamic end points. Although a drug-free period is generally desirable with aminoglycosides, this is difficult to achieve among patients with markedly impaired renal function. One also must guard against having too long a drug-free interval, although too long is not currently defined. Examples of "medication errors" could include the use of aminoglycoside in a patient with renal impairment when other, less toxic drugs are considered equally effective, or the use of high-dose, once-daily gentamicin without extending the dosing interval.

For other drugs, such as beta-lactams and vancomycin, the percent of time above the MIC (%T>MIC) should be optimized (29). In some infections, %T>MIC of at least 40 may be just as effective as %T>MIC of 100 (30). The optimal %T>MIC has been defined only in limited circumstances, and more research is needed. If the pathogen is Streptococcus pneumoniae, %T>MIC of 40 or greater may be adequate, whereas %T>MIC of 60 to 70 or greater may be needed for gram-negative pathogens (30-32).

Protein binding may also play a role in response because the observed activity in serum containing media is reduced for highly protein-bound (greater than 85%) drugs. Some investigators have suggested that activity is best characterized using free drug concentrations and exposure parameters instead of total serum concentrations (33). This principle seems more important for gram-positive organisms (e.g., Staphylococcus aureus) than for gram-negative pathogens (e.g., Pseudomonas aeruginosa and Enterobacteriaceae) (34,35). As with aminoglycosides and fluoroquinolones, doses need to be based on pharmacodynamic principles, although the principles are different. On the basis of CSF concentrations, the MIC breakpoint for penicillin susceptibility is less than 0.1 pg/mL and intermediate susceptibility is defined by a MIC of 0.1 to 1 pg/mL (27). However, patients with pneumonia due to S. pneumoniae with an MIC of 2 or 4 pg/mL may well respond to standard doses of penicillin G. The administration of 500,000 units of penicillin G per hour by continuous infusion can be expected to provide steady-state serum concentrations of about 14 pg/mL (36). Concentrations would also be expected to exceed 2 pg/mL more than 50% of a dosing interval if 2 million units were administered every 4 hours or if 3 million units were administered every 6 hours. Discussion of methods for calculating %T>MIC is beyond the scope of this chapter. Ideally, population-based pharmacokinetic parameters should be available for antibacterial drugs, and these should be determined from infected patients (target population). The parameters should be individualized based on patient specific characteristics (e.g., weight, creatinine clearance). %T>MIC can then be determined using pharmacokinetic simulation and a designated MIC value. For time-dependent antibacterial drugs, continuous infusion is the most efficient means to deliver the drug. Although there is increasing interest in continuous infusion of beta-lactam drugs and vancomycin, many situations favor the use of intermittent dosing for practical reasons (37,38). As an example, one could deliver cefepime as a continuous infusion, 125 mg/hour, or as 1 g every 8 hours, 1.5 g every 12 hours, or 3 g every 24 hours. Based on a two-compartment model and typical pharmacokinetic parameters (CLT=7.07 L/h; V1=14.3 L; (81=0.988 h-1; and (82=0.233 h-1), steady-state concentrations would remain above 8 g/mL for 100%, 68%, 55%, and 38% of the time, respectively. Achieving %T>MIC of 100 (8 g/mL) would require almost 6 g/day for every-8-hour dosing and 11 g/day for every-12-hour dosing, compared with 1.5 g/day for continuous infusion.

0 0

Post a comment