Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
Pharmacokinetic and Pharmacodynamic Basis of Optimal Antimicrobial Therapy
Acknowledgment
The authors acknowledgment contributions of Michael D. Reed to this chapter in previous editions.
As clinicians continuously strive to practice evidence-based medicine, the use of antibiotics can be frustratingly empiric. Ethical concerns, multiple confounding variables, and unclear endpoints hamper clinical infectious disease research. This empiricism is clear in most recommendations for antibiotic dosing and duration of therapy, especially in children, because only a fraction of published pharmacokinetic (PK) and pharmacodynamic (PD) data is collected from this group of patients. Although pediatric data are more plentiful and better defined for antibiotics than for other drug classes used in pediatric patients, much more work needs to be done. The lack of specific and sound dosing and safety data for drugs used in children continues to hamper the ability to determine optimal dose regimens rapidly. “Off-label” use of medications in children, unfortunately and necessarily, remains common practice. Moreover, suboptimal antibiotic dose regimens most certainly contribute to the continued emergence of the virulent, multidrug-resistant bacteria we encounter at the bedside today.
In response to this lack of pediatric data, the US Food and Drug Administration (FDA) was granted additional authority through the Best Pharmaceuticals for Children Act (BPCA, 2002, 2007, 2012) and the Pediatric Research Equity Act (PREA, 2003, 2007, 2012). The BPCA extends patent exclusivity for 6 months for subsequent research focused on a pediatric indication for on-patent drugs and prioritizes and supports pediatric labeling studies with off-patent drugs. The PREA mandates that any drug with potential use in children must also be studied in children unless it is granted a specific waiver by the FDA. This act is enforced automatically with any new drug application, including new indications for existing drugs. Failure to comply with PREA requirements can result in revocation of marketing approval for a drug. Both the BPCA and the PREA were made permanent in 2012 as part of the FDA Safety and Innovation Act (FDASIA). With a growing awareness of the important physiologic differences across the age spectrum in children and their many important influences on drug dosing, knowledge of an antibiotic’s PK and PD properties is necessary to determine optimal dosing. The increasing incidence of serious infections caused by multidrug-resistant pathogens combined with the ever-increasing emergence of virulent pathogen resistance underscores the importance of applying integrated antibiotic PK-PD–defined dosing in daily practice. , This chapter reviews basic PK principles and provides a framework for the clinician to apply these principles at the bedside.
Basic Concepts in Clinical Pharmacokinetics
Pharmacokinetics describes the time course of drug movement in the body; an understanding of a drug’s PK profile is essential to the design of an optimal dosage regimen. However, focusing on a drug’s PK profile alone provides the clinician with limited information about optimal drug dosing or, more importantly, clinical efficacy. PK principles are of clinical relevance only when they are integrated with the drug’s PD properties (i.e., the effects of the drug in the patient).
Except for intravenous (IV) drug administration, a drug must be absorbed into the systemic circulation from its site of administration. Some drugs can be administered as a prodrug requiring in vivo metabolism to liberate the active moiety. Once absorbed, the drug is distributed within the body to accessible sites that are specific to the individual drug or drug class based on inherent physicochemical characteristics (e.g., lipid solubility, molecular weight, protein binding). Factors such as age, body habitus, and disease can influence this distribution. Simultaneous with the processes of absorption and distribution, many drugs undergo metabolism before excretion from the body. A drug’s PK and disposition profile can be artificially separated into semidiscrete periods of drug absorption, distribution, metabolism, and excretion ( Fig. 291.1 ) to permit visualization and more accurate quantification. Knowledge of this information allows the clinician to predict drug concentrations and systemic exposures for the active drug that can be achieved at any time over a specified dosing interval after any dose of a given drug. With this information, an optimal drug dose and dose interval can be determined for any patient with a regimen that accounts for underlying pathophysiologic features and major organ function.
The PK property bioavailability estimates the fraction of a drug dose administered extravascularly (e.g., orally, intramuscularly) that is absorbed into the systemic circulation. Bioavailability is calculated as the area under the plasma drug concentration–time curve (AUC) achieved after extravascular drug administration divided by the AUC achieved after IV administration (e.g., AUC oral /AUC IV ). Thus, to achieve the same systemic exposure with a drug administered orally as achieved with IV dosing, one simply divides the IV dose by the drug’s bioavailability (obtained from published sources) to determine the equivalent oral dose. Lower bioavailability is not necessarily a negative attribute unless the dose required is so large that the patient tolerates it poorly. However, for orally administered antibiotics, poor bioavailability can be a significant limitation because unabsorbed drug remaining in the intestinal tract can adversely affect indigenous flora and often can lead to intestinal complaints and diarrhea (e.g., ampicillin vs. amoxicillin). Conversely, this property is sometimes therapeutically advantageous, as is the case for orally administered vancomycin to treat Clostridium difficile– associated colitis or nystatin for mucosal candidiasis.
Bioavailability depends on the molecular weight, solubility, and permeability of the drug. Highly soluble and permeable drugs have excellent bioavailability, whereas the converse results in poor bioavailability. Drugs with opposing solubility and permeability have unpredictable and variable bioavailability, even within the same patient from dose to dose because of variations in gastrointestinal (GI) pH, motility, and food content. Drugs can exhibit complex absorption patterns with delayed or multiple peaks in their blood (plasma or serum) concentrations arising from timed-release formulations, enterohepatic recycling, site-specific absorption within the GI tract, or variations in gastric emptying and intestinal transit times. A drug’s peak plasma concentration is the characteristic most significantly affected by changes in the rate and extent of absorption, and reduced peaks theoretically can compromise the efficacy of concentration-dependent antibiotics, as discussed in the later section on PD correlates for antibacterial agents.
Once absorbed into the systemic circulation, the distribution of a drug depends on specific characteristics inherent to the drug molecule. Small, non–protein-bound, nonionized lipid-soluble molecules are usually widely distributed (e.g., azithromycin), whereas larger, less lipid-soluble (i.e., polar) or highly protein-bound drugs traverse cell membranes poorly, thus restricting movement (e.g., aminoglycosides, vancomycin, echinocandins). The PK measurement that attempts to describe this process is the volume of distribution (V d ). V d is a proportionality constant that relates the amount of drug in the body to its plasma concentration. V d can be roughly conceptualized for an IV drug by the following simplified formula: V d = dose/C p , where C p is the peak plasma drug concentration. V d does not correspond to any true anatomic distribution or compartment, although the magnitude of V d provides clues about a drug’s physiologic distribution. For example, for a drug with a very small V d (i.e., 0.2 L/kg), a larger proportion of the drug is confined to extracellular fluid (composed of intravascular and interstitial fluid), whereas the distribution of a drug with a large V d can involve extensive tissue binding, intracellular distribution, or both. Knowing a drug’s V d (expressed in liters per kilogram of body weight) allows the clinician to estimate the dose necessary to achieve any desired plasma drug concentration. For example, the loading dose for a drug can be calculated from the following formula: loading dose = desired plasma drug concentration (mg/L) × V d (L/kg) × patient body weight (kg). The age-specific V d for the drug to be prescribed can be obtained from published sources or calculated for the individual patient, as outlined earlier. As inferred from the loading dose calculation, the first (or loading) dose of any drug is independent of a patient’s organ function or extent of underlying disease (e.g., renal failure). A patient’s first drug dose is dependent only on the V d of the specific drug; all subsequent doses depend on the patient’s ability to clear (eliminate or metabolize) the drug.
The PK property clearance (CL) estimates the volume of solvent (e.g., plasma, serum, or blood) from which all drug is removed per unit time. CL is therefore expressed in units of volume per time. Body CL is a composite estimate reflecting all mechanisms of drug CL, including renal, hepatic, and other forms of CL (e.g., lung), and it is calculated using the following formulas: CL = 0.693 × V d /t 1/2 , where t 1/2 is the drug’s elimination half-life; or CL = drug dose × F / AUC, where F is bioavailability. Knowledge of a drug’s CL is necessary to determine accurately the need for and proper timing of subsequent doses to maintain any desired drug concentration or degree of systemic exposure. Changes in organ function responsible for drug removal from the body are reflected by changes in a drug’s overall CL rate. Although clinically the t 1/2 often is used at the bedside as a measure of drug CL and to determine subsequent drug dosing, this value merely reflects the time required for a given drug concentration in any biologic fluid to decrease by 50%, although this is not necessarily elimination from the body, as exemplified by all forms of amphotericin. If V d is fairly constant in a patient in stable condition, t 1/2 mirrors body CL, thereby permitting bedside application. The t 1/2 can be estimated simply from the measured fall in plasma drug concentration after an individual dose (e.g., from the peak and trough plasma drug concentrations); to estimate the t 1/2 in practice, usually one must obtain 2 measured drug concentrations and calculate t 1/2 as ln(2) × −t/ln(C2/C1), where ln(2) is the natural logarithm of 2 (∼0.693), and t is the time between the 2 measured concentrations C1 and C2. Although the t 1/2 is an extremely useful bedside application of PK, the accuracy of the t 1/2 as a reflection of drug CL diminishes in situations of changing V d , renal or liver function. For example, a drug’s V d can be altered during extracorporeal membrane oxygenation (ECMO), septic shock, or severe liver disease with marked ascites.
The PK principles just outlined for V d , CL, and t 1/2 assume that the drug follows first-order or linear PK characteristics. First order means that a constant fraction of drug is cleared per unit time (i.e., regardless of the initial concentration, the time to clear x% is the same). However, for certain drugs, such as voriconazole , and phenytoin, the CL mechanisms can be saturated at clinically relevant concentrations, and thus their disposition is best described by using zero-order methodology or a combination of first-order elimination at lower concentrations and a gradual transition to zero-order elimination at higher concentrations (i.e., Michaelis-Menten kinetics) . Zero order means that a constant amount of drug is cleared per unit time (i.e., the higher the initial concentration, the longer it will take to clear x%). Fortunately, most drugs used clinically follow first-order PK, which is simpler to apply clinically. The recognition that a drug’s disposition characteristics are first order allows the clinician to use simple proportions with relative accuracy to define patient-specific drug doses. For example, if a patient’s steady-state plasma drug concentration is half that desired (at any time point), and the drug follows first-order PK, the dose can simply be doubled to achieve the desired concentration. In contrast, for a drug with zero-order characteristics, doubling the dose can result in a much higher increase in the plasma concentration.
Following repeated drug dosing, a steady-state condition is achieved (i.e., when the amount of drug administered to the patient equals the amount of drug eliminated [cleared] from the body). Simply put, “in = out.” The importance of steady-state conditions cannot be overemphasized because the drug concentrations in blood (plasma or serum) or tissue reported to be associated with drug efficacy or toxicity most often are the steady-state drug concentration. For drugs that follow first-order “proportional” PK characteristics, the drug’s t 1/2 can be used to estimate the time to reach steady-state (i.e., 4−5 times the t 1/2 to reach steady-state). A drug’s t 1/2 can be calculated for an individual patient as described earlier, or an age-appropriate t 1/2 often can be obtained from published sources and modified if necessary, based on the patient’s elimination organ function.
Unfortunately, the clinician encounters many patients who deviate from the proposed PK characteristics found in published sources, especially in pediatrics. A far more sophisticated, powerful, and nuanced approach to dosage individualization is the application of population PK modeling techniques coupled with Bayesian feedback by using specifically designed software tools. , Although these methods have the disadvantage of requiring specialized software, training, and knowledge, they have numerous advantages, including more accurate and rapid attainment of target drug concentrations without the need for steady-state conditions (including patients with unstable PK behavior), improved clinical outcomes, and reduced costs. , A complete review of this topic is beyond the scope of this chapter, but interested readers are urged to contact experts in the field and consider these approaches when desiring to optimize individual therapy.
The next sections discuss the general PK alterations that can influence both the magnitude and the frequency of antibiotic doses in neonates, patients with organ failure who are undergoing dialysis or ECMO, children with cystic fibrosis (CF), and patients with burns or septic shock. However, considerable variability exists within these generalizations, and any specific patient is likely to be best served by careful dose optimization in partnership with an expert in clinical pharmacology.
Drug Disposition in Specific Patient Populations
Neonates: the Ontogenic Basis of Drug Disposition
As children mature physically, they also mature physiologically. This principle has important and clinically relevant implications for both drug PK and PD characteristics. , Factors that influence drug absorption from the GI tract include the surface area available for absorption, pH, gastric emptying time, exocrine pancreatic function, size of the bile acid pool, and bacterial colonization ( Table 291.1 ). All these functions are variably altered in neonates, particularly those born prematurely and in young infants relative to older children and adults. , Changes in the amount and distribution of body water in neonates, as well as differences in quantitative and qualitative protein-binding characteristics, also are present and affect drug distribution. These ontogenic differences in body water composition (i.e., increased V d for water-soluble drugs) are reflected by the increased individual doses, on a milligram of drug per kilogram of body weight basis, prescribed for young infants and children compared with older children and adults. Although far from fully characterized, both the oxidative and conjugative hepatic metabolic enzyme systems are immature at birth and reach adult levels of activity at various times throughout early childhood. Renal function and elimination also mature as a function of gestational and postnatal age, with glomerular filtration reaching adult levels in infants of 34 weeks of gestation by approximately 3–5 months of age. Tubular secretion matures more slowly and reaches adult levels by approximately 8–9 months of age ( Table 291.1 ).
TABLE 291.1
Physiologic Changes in Children That Affect the Pharmacokinetic Characteristics of Drugs
Data from references 339–342.
| Parameter | Neonates | Approximate Age Approaching Adult Level |
|---|---|---|
| Absorption | ||
| Gastric pH | ↑ | 3 mo |
| Gastric emptying | ↓ | 6–8 mo |
| Pancreatic function | ↓ | 9 mo |
| Distribution | ||
| Body water | ↑ a | ∼3 yr |
| Protein binding | ↓ | 12 mo |
| Metabolism | ||
| Hepatic drug-metabolizing or hepatic drug metabolism | ↓ | 2 yr to adolescence |
| Elimination | ||
| Renal function | ↓ | Glomerular filtration: 3–5 mo |
| Tubular secretion: 8–9 mo | ||
a The distribution of body water depends on age: the total body water (TBW) of neonates is approximately 75% of body weight, with approximately 50% intracellular (IC) and 50% extracellular (EC). A gradual decrease in TBW and a shift to IC distribution occur until adult values of 50%–60% TBW, 33% EC, and 66% IC are reached at puberty.
As a result of these physiologic differences ( Table 291.1 ), infants <44 weeks of postconceptional or postmenstrual age generally have larger V d and decreased antibiotic CL, which translate clinically to lower peaks, lower overall plasma drug concentrations, and longer t 1/2 than observed in older infants and children. Furthermore, these infants can be more susceptible to drug-drug interactions at all levels, interactions that are likely to be significant for other drugs as well as antibiotics alone (see the later section on the basis for drug-drug interactions).
The physiologic differences of young infants influence not only drug PK but also PD characteristics, which necessarily incorporate the mechanism of drug action. Developmental changes in receptor function are not as relevant to antibiotics because the receptor targets are on the infecting organism. However, host receptors can be involved in therapeutic antibiotic mechanisms (e.g., human kinases that phosphorylate acyclovir) or adverse antibiotic mechanisms. Despite lower CL and prolonged t 1/2 in neonates relative to older children and adults, aminoglycosides cause less nephrotoxicity in neonates. This tolerance is thought to arise from differences in renal disposition characteristics, although the exact mechanism is unknown.
The clinician must be aware that neonates are physiologically distinct and manifest altered antibiotic PK characteristics. Dosing strategies should be evaluated critically to make prescribing decisions to optimize therapeutic success. (See Appendix 292.1 for dosing guidance.) In the absence of neonatal dosing recommendations, knowledge of the altered PK patterns will make empiric dosing more rational when it is extrapolated from dosing recommendations for older children or adults.
Patients With Organ Failure and Principles of Dialyzable Drugs
The major routes of drug elimination are by the kidneys into urine and by the liver into bile. For most drugs, only minor amounts are eliminated into other body fluids. Alterations in drug disposition result from failure of either of these organs of elimination or failure of the cardiopulmonary system to support their proper function. Patients in cardiopulmonary failure can experience alterations both in V d as a result of increased total body water and in CL as a result of hepatic or renal dysfunction. Knowledge of the PK properties of individual drugs is crucial to understanding the magnitude of such effects. Drugs with a small V d are more greatly affected because they tend to be distributed to extracellular fluid. The route of metabolism and elimination determines the impact of hepatic or renal dysfunction.
Extracorporeal Membrane Oxygenation
ECMO influences drug disposition in several ways. ECMO circuitry can add up to 500 mL of volume to the patient, in which case V d is increased; children weighing <20 kg will most likely experience corresponding lower peak drug concentrations and a longer t 1/2 . Further, drugs can bind to the oxygenation membrane, thereby potentially increasing both V d and CL. Binding is increased with increasingly lipophilic drugs. CL can be reduced because of altered renal and hepatic perfusion. ,
Alterations in the PK characteristics of antibiotics during ECMO have been relatively well characterized for gentamicin and vancomycin. , Although the V d of aminoglycosides is variably increased, the t 1/2 can be doubled. Neither drug significantly binds to the oxygenation membrane, as predicted by low lipophilicity. The implication for dosing is that aminoglycosides should be administered every 18–24 hours initially (or less frequently in patients with renal failure) to infants undergoing ECMO (with monitoring of serum drug concentrations), although no efficacy studies are available to support this recommendation. For vancomycin, V d and t 1/2 are increased, and CL is decreased, resulting in a suggested initial dose of 20 mg/kg IV every 18–24 hours for infants with a serum creatinine value <1.5 mg/dL. For infants with a higher serum creatinine concentration, the dosing interval must be extended. In all cases, serum vancomycin concentrations should be monitored.
Increased ECMO-associated cefotaxime V d has been documented, but the percentage of the dosing interval that is greater than the organism’s minimal inhibitor concentration (MIC) after standard doses during ECMO is not significantly different from the same doses while the patient is not undergoing ECMO, thus leading to no required dosing changes. Some reports have noted lowered plasma voriconazole concentrations requiring higher doses in patients receiving ECMO, a finding consistent with its high lipophilicity, and therapeutic drug monitoring is recommended. , Caspofungin, which is less lipophilic, had reportedly lowered and unaltered concentrations in 3 patients receiving ECMO, with the result that no clear recommendation was made. Oseltamivir PK does not appear to be changed significantly. Indeed, a case report of a 6-year-old boy receiving both ECMO and continuous veno-venous hemofiltration (HF) demonstrated that low plasma levels of oseltamivir were likely caused by poor oral absorption rather than by marked changes in V d or CL.
For other antimicrobial agents for which plasma drug concentrations are unavailable, the clinician can only estimate dosing modifications based on the expected changes described earlier. , In general, V d and t 1/2 are the same or are increased. Lipophilic drugs are more greatly affected than are hydrophilic drugs. However, patients undergoing ECMO usually have multiorgan failure, with renal failure and the use of continuous veno-venous HF the most problematic for many antibiotic regimens.
Dialysis
For patients in acute or chronic renal failure who require dialysis, several options can be used for extracorporeal therapy, including intermittent or continuous hemodialysis (HD), HF, or peritoneal dialysis (PD). HD and HF therapies can be combined. , HD removes an antibiotic by diffusion across the dialysis membrane into the dialysate according to the drug’s concentration gradient. The amount of blood flow past the dialysis membrane and the composition and flow of dialysate influence the amount of drug removed. HF (without HD) removes antibiotics in the ultrafiltrate, which is a hydrostatically generated flow of water containing the dissolved drug. No drug concentration gradient is present. PD, in general, removes most antibiotics minimally. The volume of PD fluid generally is small (0.05–0.2 L/kg) in comparison with the V d of antibiotics (0.2 to >100 L/kg), so only a very small percentage of the total drug in the body is distributed to the PD fluid unless the antibiotic has a similarly small V d . Furthermore, the intraperitoneal concentration of antibiotics is proportional to the concentration of free, non–protein-bound drug. The protein content of typical dialysate fluid is 0, so highly protein-bound antibiotics are poorly distributed to the peritoneal fluid. Increasing the amount of PD dialysate or the frequency of exchange increases the amount of drug extracted, but a cumulative amount extracted of <25% during a dosing interval is usually insignificant. In contrast, antibiotics administered into the peritoneal dialysate can reach therapeutic plasma concentrations if they can cross the peritoneal membrane, passage through which is enhanced in the presence of inflammation. Because the potential extra-abdominal volume for an antibiotic to diffuse into is much larger than the intra-abdominal volume, with equal concentrations across the peritoneal membrane at equilibrium, the amount of drug in the body is greater than in the peritoneal dialysate.
All these modalities partially restore the ability of the body to eliminate an antibiotic that normally would be cleared by the kidneys. Although several thorough reviews have described the use of antimicrobial agents in these patients, none of the guidelines has pediatric dosing, and all strongly convey the same message: antimicrobial concentrations highly depend on factors intrinsic to the mode of dialysis itself and the drug’s PK properties ( Table 291.2 ). This variability could have significant clinical implications. , Although the clinician can apply principles outlined here to an individual patient to generate a qualitative assessment of the need for dose adjustment, confirmatory measurement of drug concentrations is crucial when possible.
TABLE 291.2
Factors That Increase the Likelihood of Antibiotic Removal by Dialysis
| Factors Intrinsic to the Antibiotic | Factors Intrinsic to the Mode of Dialysis |
|---|---|
| Smaller volume of distribution (<1 L/kg) | CVVHDF a > CVVH a > CVVHD a > IHD ≫ PD |
| Smaller molecular mass (<2 kd) | Frequent dialysis |
| Reduced protein binding (<80%) | High flow rate of blood or dialysate |
| Neutral charge | Filter capacity and selectivity |
| Normally eliminated by kidneys (>30%) | Dialysate composition |
CVVH, continuous venovenous hemofiltration; CVVHD, continuous venovenous hemodialysis; CVVHDF, continuous venovenous hemodiafiltration; IHD, intermittent hemodialysis; PD, peritoneal dialysis.
a The differences in drug clearance among the continuous modes of hemodialysis or hemofiltration are less significant for small molecules, which include most hydrophilic antibiotics that are likely to be cleared by these modes. Among these, the largest are teicoplanin (1.9 kd), colistimethate, also known as colistin (1.8 kd), and vancomycin (1.5 kd).
Drugs with a small V d (<1.0 L/kg) that are primarily confined to the extracellular compartment are available for filtration across the dialysis membrane. The pore size of HD membranes is approximately 0.5 kd, whereas HF membranes are usually approximately 50 kd. Most antibiotics are <1 kd. Passage across a membrane is inversely proportional to the square root of the molecular mass. Drug-protein complexes are too large to pass through the pores of intact dialysis membranes, so highly protein-bound antibiotics (>80%) pass through the membrane poorly. Further, the membranes carry a net negative charge, and highly polar molecules do not cross the membrane efficiently. Antibiotics that are mainly eliminated by a functioning kidney are more likely to be eliminated by dialysis, and a simplistic rule of thumb predicts some removal of drugs that are eliminated >30% unchanged in the urine in patients with normal renal function. However, the fraction of an antibiotic normally cleared by active tubular secretion is not compensated for by dialysis, and this factor should be considered in any patient manifesting unexplained signs or symptoms of drug toxicity.
Antibiotic dose adjustment in the absence of measurable drug concentrations in blood, dialysate, or ultrafiltrate is a crude estimate at best. Unfortunately, most pharmacology references consulted to determine dose recommendations use complicated formulas that require information the clinician often does not possess. Therefore, a more intuitive approach is typical, and the clinician should first consider the relevance of potential drug accumulation by monitoring the patient closely for desired and undesired drug effects. Fortunately, most antibiotics have a very high therapeutic index, i.e., the ratio of therapeutic to toxic doses. “Safe” drugs can generally be allowed to accumulate moderately without risk of serious or irreversible toxicity, or both. In contrast, more rigorous dose adjustments should be applied for antibiotics that can be associated with toxicity (e.g., the aminoglycosides) because patients with renal failure, especially if uremic, generally are more prone to adverse effects from all drugs. Regardless of the dosing scheme selected, the clinician must monitor patients closely for specific evidence of toxicity related to the drug in question as well as expected clinical responses (see Chapter 289 ). Conversely, unaccounted-for extracorporeal elimination has major importance when subtherapeutic drug concentrations lead to clinical failure, which can happen when a drug dose is based on a patient’s intrinsic renal function without accounting for dialysis-mediated drug CL. Therefore, for most antibiotics, it is more important to avoid underdosing. Appendix 291.1 lists suggested adjustments for selected antibiotics in response to renal failure with or without dialysis.
You're Reading a Preview
Become a Clinical Tree membership for Full access and enjoy Unlimited articles
If you are a member. Log in here