Principles of Anti-infective Therapy

George M. Eliopoulos, Robert C. Moellering Jr. *

Keywords

antimicrobial agent selection; antimicrobial combinations; host factors; organism factors; risks of therapy

Although the discovery of effective agents to prevent and treat infection caused by bacteria and other pathogenic microorganisms is one of the most important developments of modern medicine, the use of such agents is not limited to the present era. Substances with anti-infective potential have been applied medically for thousands of years. Indeed, more than 2500 years ago, the Chinese were aware of the therapeutic properties of moldy soybean curd applied to carbuncles, boils, and other infections,¹ and ancient Greek physicians, including Hippocrates, routinely used substances with antimicrobial activity, including wine, myrrh, and inorganic salts, in their treatment of wounds.² Until the discovery of the microbiologic basis of infections in the 19th century, however, the therapy for infections remained strictly empirical. Heavy metals, such as arsenic and bismuth, were found to be useful against a number of infections, including syphilis, in the early 1900s, but the modern era of chemotherapy did not really begin until the discovery and initial clinical use of the sulfonamides in 1936.¹ This was followed in the 1940s by the discovery of the therapeutic value of penicillin and streptomycin, and by 1950 the golden age of antimicrobial chemotherapy was well under way.

It is the relatively recent work in this area (since 1936) that forms the basis for this and each of the succeeding chapters on anti-infective therapy. The major emphasis in this chapter is on antibacterial agents because more data are available on these drugs. However, many of the principles to be discussed can also be applied to the use of antifungal, antiviral, and, to some extent, antiparasitic drugs.

Choice of the Proper Antimicrobial Agent

In choosing the appropriate antimicrobial agent for therapy for a given infection, a number of factors must be considered. First, the identity of the infecting organism must be known or, at the very least, it must be possible to arrive at a statistically reasonable guess as to its identity on the basis of clinical information. Second, information about the susceptibility of the infecting organism, or likely susceptibility, must be as accurate as possible. Finally, a series of factors specific to the patient who is being treated must be considered to arrive at the optimal choice of antimicrobial agent. Each of these items is considered in this section.

Identification of the Infecting Organism

Several methods for the rapid identification of pathogenic bacteria in clinical specimens are available. A Gram-stain preparation is perhaps the simplest and least expensive method to determine the presence of bacterial and some fungal pathogens. This technique can be used to visualize microorganisms in body fluids that are normally sterile (cerebrospinal fluid, pleural fluid, synovial fluid, peritoneal fluid, urine) and, by their morphologic features, to assign them to broad categories against which specific antimicrobial agents are likely to be effective. Preparations of sputum may also be helpful in revealing the nature of the infecting organism in patients with bacterial pneumonia.

Although infrequently done at present, a Gram stain of a stool specimen may produce useful information. Polymorphonuclear leukocytes are not found in normal stools. When present, they suggest the possibility of a bacterial gastroenteritis, such as shigellosis, salmonellosis, campylobacteriosis, or invasive Escherichia coli gastroenteritis. Polymorphonuclear leukocytes are not found in the stools of patients with viral gastroenteritis, food poisoning, cholera, and diarrhea caused by noninvasive toxigenic E. coli.³ Campylobacter may be identified in the stools of patients by its characteristic gull-wing appearance on smear.⁴

Immunologic methods for antigen detection, such as enzyme-linked immunosorbent assay (ELISA) or latex agglutination, may also provide clues for the rapid identification of the infecting pathogens. Molecular techniques are increasingly being applied to the detection and identification of microbial agents. The polymerase chain reaction (PCR) has been used to identify RNA or DNA of viruses, bacteria, and other microorganisms in the blood and body fluids of patients,⁵^,⁶ and this technique and others, including the use of DNA probes, have proved to be helpful in rapid identification of organisms that have been cultured in the laboratory or that are present at high density in direct clinical specimens.⁷ Final and definitive identification of pathogenic organisms still usually requires culture technique. Furthermore, molecular systems that allow determination of antimicrobial susceptibility as well as identification are feasible, but are only now being introduced into clinical laboratories.

Recent initiatives to reduce the burden of hospital infections caused by methicillin-resistant Staphylococcus aureus (MRSA) and other multidrug-resistant bacteria through active surveillance have led to increased use of molecular methods for rapid detection of these organisms.⁸ Nucleic acid amplification tests have now become widely available to detect Neisseria gonorrhoeae and Chlamydia trachomatis from various body sites. The disadvantage of these systems at present is that they do not provide susceptibility information; thus, culture for susceptibility testing is essential when the patient has not responded to recommended therapy for gonococcal infection. Molecular methods for the detection of Mycobacterium tuberculosis are now available that also determine susceptibility of the organism to rifampin.⁹ Nevertheless, in almost all cases where bacteria are identified by molecular methods, when treating infections it is imperative that appropriate specimens be obtained for culture before beginning antimicrobial therapy, to test for susceptibility to agents that may be used in therapy. Once anti-infective therapy has been started, cultures often are rendered sterile, even though viable organisms remain in the host. One example where antibiotic therapy would be initiated before cultures are obtained is in suspected bacterial meningitis in circumstances where lumbar puncture cannot be obtained promptly. In such cases, blood cultures should be obtained before antibiotic therapy, and cerebrospinal fluid should be sampled as soon as it is feasible to do so.

In most cases, it is impossible to determine the exact nature of the infecting organisms before the institution of antimicrobial therapy. In these cases, the use of bacteriologic statistics may be particularly helpful.¹⁰^,¹¹ The term bacteriologic statistics refers to the application of knowledge of the organisms most likely to cause infection in a given clinical setting. For example, a person with normal host defense mechanisms who develops cellulitis of the arm after a minor abrasion most likely has an infection caused by Streptococcus pyogenes, other β-hemolytic streptococci, or possibly Staphylococcus aureus, and antimicrobial therapy should be tailored accordingly, even though there is no material available for culture or examination with Gram stain. Similarly, a young child with acute otitis media almost certainly has an infection caused by either a virus or one of four major bacterial pathogens: Haemophilus influenzae, Streptococcus pneumoniae, Moraxella catarrhalis, or a group A streptococcus.¹²

Determination of Antimicrobial Susceptibility of Infecting Organisms

Because different organisms vary in their susceptibility to antimicrobial agents, it is imperative that we have a way to determine the antimicrobial susceptibility of the actual (or presumed) infecting organism or organisms. If the pathogen is isolated from a culture, it can be subjected to direct susceptibility testing, as described in Chapter 16. A number of methods for determining antimicrobial susceptibility are available. A simple and still widely used method is the disk diffusion test, which determines susceptibility of an organism based upon the zone of inhibition of the organism seeded onto an agar plate by using radial diffusion of an antimicrobial from a paper disk impregnated with the drug. A refinement of the disk diffusion technique uses antimicrobial gradient strips (e.g., Etest by bioMérieux [Marcy-l’Etoile, France]; M.I.C.E. [minimal inhibitory concentration evaluator] by Oxoid [Cambridge, England]) applied to agar plates seeded with the test organism. With these methods, intersection of the inhibition zone with the graduated strip permits determination of an actual minimal inhibitory concentration end point.¹³

Quantitative data are also provided by methods that incorporate serial dilutions of antimicrobials in agar-containing or broth culture media. The lowest concentration of the antimicrobial agent that prevents visible growth, usually after an 18- to 24-hour incubation period, is the minimal inhibitory concentration (MIC). The minimal bactericidal concentration (MBC), or minimal lethal concentration (MLC) may be determined in broth dilution tests by subculturing the containers that show no growth onto antibiotic-free agar-containing media. The lowest concentration of antimicrobial that totally suppresses growth on antibiotic-free media (or results in a 99.9% or greater decline in colony count) after overnight incubation is known as the MBC (or MLC). The aforementioned techniques are based on an 18- to 24-hour incubation period. A variety of rapid methods are available as well.¹⁴ These are based on a determination of changes in bacterial growth rates caused by antimicrobial agents and can provide susceptibility information in 4 to 8 hours.

Susceptibility testing is particularly important for certain organisms, such as S. aureus and the various facultative and aerobic gram-negative bacilli. The widespread clinical and agricultural use of antimicrobials since the 1930s and 1940s has resulted in the emergence of many strains of bacteria that are resistant to one or more antimicrobial agents (see Chapter 18).¹⁵^–¹⁷ In most cases in which adequate studies have been done, it appears that the role of antimicrobial agents is to exert selective pressure that results in the emergence of resistant organisms. In some cases, the organisms are naturally resistant to the antibiotic used. Examples of the latter include gram-positive organisms, such as staphylococci and streptococci, that are naturally resistant to aztreonam and the polymyxins. Most gram-negative bacilli are naturally resistant to penicillin G, erythromycin, and clindamycin.

In other cases, the resistant bacterial strains have acquired genes encoded on transposons or plasmids that enable them to resist antimicrobial inhibition. These genes may provide the organisms with the ability to synthesize enzymes that modify or inactivate the antimicrobial agent, they may result in changes in the bacterial cell’s ability to accumulate the antimicrobial agent; they may permit the cell to produce alternative metabolic enzymes resistant to inhibition by the antimicrobial agent, or they may modify the target so that it becomes resistant to the action of the antimicrobial.¹⁵ Examples of each of these mechanisms of resistance are well known. Most strains of S. aureus are resistant to penicillin and contain plasmids that enable them to produce an extracellular β-lactamase that hydrolyzes and inactivates penicillin G.¹⁵ Many gram-negative bacilli that are resistant to aminoglycosides, such as streptomycin, tobramycin, gentamicin, and amikacin, contain genes on plasmids that code for the production of periplasmic enzymes that catalyze a modification of the aminoglycosides by phosphorylation, acetylation, or adenylation.¹⁵ Efflux mechanisms, which may be plasmid or transposon mediated, in S. aureus, pneumococci, and gram-negative bacilli, can cause resistance to tetracycline, the macrolides, and other agents.¹⁸^,¹⁹ E. coli resistant to trimethoprim has been found to contain genes that enable it to synthesize a new dihydrofolate reductase (the enzyme specifically inhibited by trimethoprim) that is 10,000 times less susceptible to the in vitro effects of trimethoprim than is the host bacteria’s own chromosomal enzyme.¹⁵ Other extrachromosomal resistance genes may confer resistance to aminoglycosides or to various other protein synthesis inhibitors, including linezolid, chloramphenicol, and clindamycin, by methylation at specific positions on 16S ribosomal RNA (rRNA) or 23S rRNA, respectively.²⁰^,²¹

These developments provide the rationale for performing tests of antimicrobial susceptibility whenever there is reasonable doubt about the susceptibility of a given organism that is thought to be pathogenic. There are very few examples of organism-antibiotic combinations for which susceptibility can be predicted with a sufficiently high degree of certainty that testing would be unnecessary in the setting of a severe infection. Group A and other β-hemolytic streptococci remain susceptible to the penicillins and cephalosporins. Thus, testing these organisms against these particular agents need not be carried out routinely at the present time. However, even a statement such as this is fraught with a certain amount of danger. Isolates of group B streptococci have now been encountered with reduced susceptibility to β-lactam antibiotics.²² The discoveries of penicillin-resistant meningococci; the emergence of fluoroquinolone-resistant gonococci in Asia, Africa, and the United States; the rapid spread of ampicillin-resistant strains of H. influenzae; the proliferation of vancomycin-resistant enterococci; the emergence of vancomycin-intermediate and vancomycin-resistant staphylococci; and the discovery of streptomycin-resistant Yersinia pestis make us realize that, in time, strains of virtually any organism may be found that are resistant to antimicrobial agents that previously had been effective against them.²³^–²⁷

It is important to consider geographic differences in patterns of susceptibility of organisms when choosing antimicrobial agents. In many cases, there may be variations in susceptibility patterns between hospitals and the community, between neighboring hospitals, or even among units within a single hospital. For example, for many years, cases of MRSA infection were almost always associated with hospitals and other health care facilities, but that has changed, and community-acquired MRSA infections in persons who have had little or no contact with health care systems are common.²⁸^,²⁹ In addition, community-associated strains have now entered hospitals, where they can account for a substantial proportion of hospital-onset MRSA infections.³⁰

All of these facts must be considered when choosing initial therapy for various infections. Tables listing drugs of choice, such as those in the Treatment Guidelines from the Medical Letter, must be updated frequently to keep up with changes in antimicrobial resistance patterns.³¹^,³² Because resistance rates can vary among facilities, periodic updates of institutional cumulative susceptibility reports by organism and antimicrobial agent can provide valuable local information to the local health care practitioner.³³

Host Factors

It is clearly important to determine the identity and antimicrobial susceptibility of the organism or organisms causing a given infection. However, optimal therapy is impossible unless we also consider a number of host factors that may influence the efficacy and toxicity of antimicrobial agents.³⁴

History of Previous Adverse Reactions to Antimicrobial Agents

Simply obtaining an adequate history of previous adverse reactions to drugs may prevent the inadvertent administration of an antimicrobial agent to which the patient is allergic or otherwise intolerant. A failure to do so can have serious (and sometimes fatal) consequences (see Chapter 23).

Age

The age of the patient is a major factor in choosing an antimicrobial agent. Gastric acidity varies with age. The pH of gastric secretions is higher in young children and does not reach adult levels of acidity until approximately the age of 3 years. At the other end of the age spectrum, there can also be a decline in gastric acidity. Early studies reported gastric achlorhydria in 5.3% of people 20 to 29 years of age, in 16% of those 40 to 49 years, and in 35.4% of those older than 60 years.³⁴ More recent work suggests that 90% of individuals older than 60 years can acidify gastric contents.³⁵ Nevertheless, many people of all age groups routinely take agents that interfere with normal gastric acid production. The absorption of a number of antimicrobials administered via the oral route depends on their acid stability and the pH of gastric secretions. Penicillin G is an excellent example of this phenomenon. The oral absorption of penicillin G is markedly reduced by gastric acid. However, in young children and in achlorhydric patients, the absorption of the drug is markedly enhanced. As a result, various orally administered penicillins produce high serum levels in young children and in older patients who have achlorhydria. The absorption of other orally administered β-lactam antibiotics is probably also enhanced in achlorhydric patients; however, evidence is convincing only in the case of the penicillins.³⁶ Gastric acidity does not always have a negative influence on the absorption of antimicrobials. For example, cefpodoxime, administered as the β-lactam prodrug cefpodoxime proxetil, appears to be better absorbed at a low pH.³⁷

Renal function, likewise, varies with age. It is relatively diminished in premature and newborn children and reaches adult levels between 2 and 12 months of age.³⁴ Thus, the serum half-lives of drugs that are primarily excreted by the kidneys may be considerably increased in neonates. As a result, doses of antimicrobial agents such as penicillin G and its various semisynthetic derivatives, as well as the aminoglycosides, must be altered in neonates.

Aging results in the decline of a number of physiologic processes, including renal function.³⁶ Creatinine clearance may be significantly reduced in older patients even though they have normal blood urea nitrogen or serum creatinine concentrations. In view of this, high doses of the penicillins or cephalosporins or carbapenems must be given with caution to older adults to prevent the development of excessively high serum levels, which may produce severe neurotoxic reactions, such as myoclonus, seizures, and coma.³⁴^,³⁶ Other adverse reactions to the penicillins, such as reversible neutropenia, may be dose related and may occur with increased frequency when high doses of such drugs are given to older patients with unrecognized renal impairment.³⁶ This, however, has not been proven. Impaired renal excretion of the aminoglycoside antibiotics may result in elevated serum concentrations, which in turn may be associated with an increasing incidence of ototoxicity in older patients.³⁸

In addition to the toxicity that may result from impaired renal excretion in neonates and older adult patients, other adverse effects of antimicrobial agents may also be age related.³⁶^,³⁹ Hepatic function in the neonate is underdeveloped by adult standards. This can result in difficulties if such patients are administered drugs that are normally excreted or inactivated by the liver.

The sulfonamides compete with bilirubin for binding sites on serum albumin. When given to neonates, they produce increased serum levels of unbound bilirubin that predispose the child to kernicterus.⁴⁰^,⁴¹

The tetracyclines are avidly bound to developing bone and tooth structures. As they bind to developing teeth, they may cause a number of adverse effects ranging from purplish to brownish discoloration of the teeth to actual enamel hypoplasia.³⁴^,⁴⁰ The tetracyclines readily cross the placenta.⁴² Thus, when administered during the latter half of pregnancy or from birth to the age of 6 months, they may cause these effects on the deciduous teeth of the infant. From the age of 6 months to 6 to 8 years, similar damage to the permanent teeth may occur. In view of this, tetracyclines should be avoided, if possible, in young children.

The quinolone antimicrobials have been shown to cause cartilage damage and arthropathy in young animals. As a result, they had not been recommended for use in children.⁴³ More recent evidence, however, suggests that the risk of arthropathy is less than previously thought, opening the way for judicious use of fluoroquinolones in children⁴⁴; both ciprofloxacin and levofloxacin have a U.S. Food and Drug Administration (FDA) indication in children for prophylaxis after inhalational exposure to Bacillus anthracis. Nevertheless, when alternatives with more established track records in children are not available, these drugs should be used judiciously where indicated because of additional concerns relating to the emergence of resistance.⁴⁵

Adverse effects caused by a number of antimicrobial agents have been noted to occur with increased incidence in older adults.³⁶ In some cases (and perhaps in all if adequately studied), this relationship may be shown to be caused by specific disease states or by impairment of physiologic processes associated with aging, as noted earlier. However, in certain cases, no specific factors other than age can be identified. The hepatotoxicity associated with isoniazid administration is a good example of this. A small percentage of patients receiving isoniazid develop toxic hepatitis that may be fatal if not recognized in time.⁴⁶ Liver damage from isoniazid is rare in children but can be severe.⁴⁷ In patients 20 to 34 years of age, the incidence of isoniazid hepatotoxicity is 0.3%, and it rises steadily with age to reach 2.3% in patients 50 years or older.⁴⁸ Despite these data, isoniazid prophylaxis is still thought to have a positive risk-benefit ratio in older patients, as long as liver function tests are monitored.⁴⁹^,⁵⁰

Hypersensitivity reactions to antimicrobial agents also appear to be more common in older adults than in younger patients.³⁴ This seems to be because older patients are more likely to have been previously exposed, and thus sensitized, to these agents.

Genetic or Metabolic Abnormalities

The presence of genetic or metabolic abnormalities may also have a significant effect on the toxicity of a given antimicrobial agent. In a small proportion of individuals treated with the antiretroviral drug abacavir, a severe hypersensitivity reaction can occur, consisting of fever, rash, and abdominal and respiratory symptoms. The presence of a human leukocyte antigen allele, HLA-B*5701, has been found to be highly associated with immunologically confirmed cases of abacavir hypersensitivity reaction.⁵¹ It has now become common practice to screen for the presence of this allele before initiating therapy with abacavir. More recently, the presence of HLA-B*13:01 was found to be associated with risk of developing dapsone hypersensitivity syndrome; absence of the allele reduced the risk sevenfold.⁵² The genetically determined metabolism and clearance of a number of drugs may affect the toxicities that are encountered with their use. Data suggest that genetically determined activities in at least three enzymes that are involved in the disposition of isoniazid and its metabolites—N-acetyl-transferase 2, the cytochrome P-450 enzyme CYP2E1, and glutathione-S-transferase—are likely to influence the risk of hepatotoxicity with this agent during treatment of tuberculosis.⁵³

A number of antimicrobial agents have been shown to be capable of provoking hemolysis in individuals with glucose-6-phosphate dehydrogenase (G6PD) deficiency, including the sulfonamides, dapsone (a sulfone), nitrofurantoin, and various antimalarial compounds.⁵⁴ Sulfonamides may likewise cause hemolytic reactions in the presence of certain hemoglobinopathies, including hemoglobin Zurich and hemoglobin H.³⁴

The presence of metabolic disorders, such as diabetes mellitus, may also pose problems in antimicrobial therapy. Certain agents, such as the sulfonamides (especially the long-acting types), can potentiate the hypoglycemic activity of sulfonylurea hypoglycemic agents.³⁶ Agents of the fluoroquinolone class have been associated with dysglycemic reactions, both hypoglycemia and hyperglycemia. Individuals with baseline glucose abnormalities and those receiving treatment for diabetes may be particularly at risk.⁵⁵^,⁵⁶ The dextrose load infused with intravenous antibiotics dissolved in dextrose-containing vehicles may be sufficient to produce hyperglycemia and glucosuria in diabetic patients. In the past, another kind of “glucosuria” could occur in patients receiving antimicrobial agents. The cephalosporins, isoniazid, nitrofurantoin, penicillin, streptomycin, and the tetracyclines can all cause false-positive test results when urine sugar levels are determined by a method that measures reducing substances in the urine (e.g., the Benedict test or Clinitest).³⁶ Tests that are specific for glucose (i.e., that use glucose oxidase) are not affected by antimicrobial agents.⁵⁷ Patients with diabetes mellitus represent a group that may also be particularly vulnerable to another uncommon but well-documented adverse effect associated with fluoroquinolone therapy—tendon rupture.⁵⁵

The concomitant administration of chloramphenicol has been noted to delay the reticulocyte response to vitamin B₁₂ or iron therapy in patients with pernicious anemia or iron deficiency anemia.³⁴ Patients with pernicious anemia and gastric achlorhydria may exhibit enhanced serum levels of antimicrobials, such as penicillin G, when given by the oral route.

Rifampin and other rifamycins may increase the hepatic metabolism and therefore decrease the effect of oral anticoagulants, oral contraceptives, barbiturates, and a number of other drugs, including the protease inhibitors.⁵⁸ It is equally important to recall this interaction when discontinuing rifamycin therapy in patients whose dose of a concomitant agent had been adjusted upward during rifampin treatment; the dose of the other agent is likely to require readjustment after discontinuing the rifamycin, to avoid undesirably high serum levels and adverse effects (e.g., bleeding from overanticoagulation with warfarin).

Drugs from several antimicrobial classes, including macrolides, fluoroquinolones, and antifungals, can potentially prolong the cardiac QTc interval (www.azcert.org). In addition, inhibition of cytochrome P-450 drug elimination pathways by some members of these antimicrobial classes can increase plasma concentrations of nonantimicrobial agents that have even more significant effects on cardiac repolarization.⁵⁹ Individuals with underlying conduction abnormalities or those receiving other medications that can affect cardiac electrical activity may be particularly vulnerable to arrhythmias (torsades de pointes) related to antimicrobial use, but care should be taken to weight the potential risks and benefits when prescribing such agents to any individual who is at risk of cardiovascular events.⁵⁹^,⁶⁰

Pregnancy

Patients who are pregnant and nursing mothers also pose certain challenges in the selection of appropriate antimicrobial agents. All antimicrobial agents cross the placenta in varying degrees.⁶¹^,⁶² Thus, the use of such agents in pregnant women provides direct exposure of the fetus to adverse effects of the drug. Because of the complexity of assessing factors influencing the safety and efficacy of antibiotics in pregnancy, the selection of appropriate agents is best undertaken with expert guidance. Although there are few solid data on the teratogenic potential of most antimicrobial agents in humans, experience suggests that certain drugs, such as the penicillins, the cephalosporins, and erythromycin, and antituberculous drugs (such as isoniazid, rifampicin, and ethambutol), are unlikely to be teratogenic and are safe for pregnant women to use with appropriate care.⁴⁰^,⁶¹^–⁶³ Several antimicrobial agents are known to have teratogenic potential. Quinine, ribavirin, and miltefosine are listed as FDA pregnancy risk category X, indicating that because of known human reproductive effects, the benefits are outweighed by risks.³² Other agents, such as tetracyclines (including tigecycline), aminoglycosides, efavirenz, and voriconazole, are listed as pregnancy category D drugs, indicating that although there is evidence of human risk, there are situations in which the benefits may outweigh those risks.³² Evidence pertaining to the safe use of common antimicrobials in pregnancy has been reviewed.⁶⁴^,⁶⁵

The possible adverse effects of tetracyclines on fetal dentition have already been noted. In addition, pregnant women receiving tetracycline are particularly vulnerable to certain toxic effects, including acute fatty necrosis of the liver, pancreatitis, and probably renal damage.³⁴ The liver damage may be severe and can result in death. When administered to patients with impaired renal function, these effects may be magnified, particularly if the agent is one of the tetracyclines that are primarily excreted by the kidneys. These adverse effects are dose related and may be more frequent after intravenous administration.

The aminoglycosides cross the placenta. Fetal eighth nerve toxicity has been reported in infants of mothers exposed to prolonged courses of streptomycin or kanamycin.⁶⁵^,⁶⁶ Because hearing loss or vestibular dysfunction can be side effects of other aminoglycosides in patients receiving these agents, the other aminoglycosides are also labeled as pregnancy category D.³²^,⁶⁷

Another aspect of drug therapy in pregnancy is that the pharmacokinetic behavior of the drug may be altered, particularly in late pregnancy. For example, the pharmacokinetic behavior of antiretroviral agents may be influenced by a number of factors related to absorption, distribution, and elimination of the various agents.⁶⁸ It has also been found that serum levels after a given dose of ampicillin are lower in pregnant than in nonpregnant women.⁶⁹ This is related to more rapid clearance of the drug and to a greater volume of distribution (probably resulting from increased plasma volume) in pregnancy. Thus, higher doses of ampicillin may be required to achieve therapeutic blood levels in pregnancy. It is likely that these observations also apply to other antimicrobial agents, but data on these are limited. For the antiretroviral protease inhibitor lopinavir/ritonavir, plasma concentrations are lower in women in the second and third trimesters of pregnancy than in nonpregnant women, and increased doses and/or monitoring of drug concentrations may be necessary.⁷⁰

The clinician should note that early exposure to antibiotics, either from peripartum administration to the mother or for treatment of infection in the child, has the potential to perturb natural development of the newborn’s intestinal microbiome.⁷¹ Conceivably, such changes could lead to downstream effects on immune response or adipokine regulation.⁷²

Virtually all antimicrobial agents appear in measurable concentrations in breast milk when administered in therapeutic doses to nursing women.⁷³ The amount of drug excreted into breast milk depends on its degree of ionization, its molecular weight, and its solubility in fat and water. Under usual circumstances, the concentrations of antibiotics found in breast milk are quite low. However, even these small amounts may cause significant adverse reactions in the nursing infant. In a mouse model, exposure of mice at the time of weaning to subtherapeutic concentrations of antibiotics led to changes in the colonic microbiome and adiposity.⁷⁴ Nalidixic acid and the sulfonamides in breast milk have been shown to cause hemolysis in infants with G6PD deficiency. Sulfonamides in breast milk may be dangerous to premature babies because even small doses of ingested sulfonamides may produce increased levels of unbound bilirubin by displacing bilirubin from its albumin-binding sites. As noted previously, this predisposes the child to kernicterus.⁷³ The possibility that antimicrobial agents in breast milk can sensitize newborn children is a theoretical one and has not been convincingly demonstrated. Although tetracycline is excreted in breast milk, it is unlikely to produce damage to the nursing child’s bones or teeth because the calcium in the milk forms an insoluble chelate with tetracyclines, and the chelate is not absorbable by the oral route.⁷³

Renal and Hepatic Function

The ability of the patient to metabolize or excrete antimicrobial agents is one of the most important host factors to consider, especially when high serum or tissue concentrations of the administered drugs are potentially toxic. From a practical point of view, this means that the clinician must assess the patient’s renal and hepatic function carefully because these organs serve as the major, and in most cases the only, routes of excretion and inactivation of antimicrobials. Renal excretion is the most important route of elimination of most antimicrobial agents.⁷⁵^–⁸⁰ Doses for drugs that require alteration in patients with impaired renal function can be found in the chapters dealing with the individual agents and in Chapter 54. In general, agents that require no dosage change in patients with impaired renal function are excreted effectively by extrarenal routes, usually the hepatobiliary system, in patients with renal failure. Their use in normal doses does not result in the appearance of toxic serum levels in this situation, although the urine levels of a number of these agents (e.g., doxycycline, moxifloxacin) may be diminished significantly.

Toxic serum levels of certain agents may develop if they are used without dosage modification in patients with impaired renal function. Excessive serum levels of penicillin G or imipenem may be associated with neuromuscular hyperexcitability, myoclonus, seizures, or coma.³⁴ Similar effects may be seen at toxic concentrations of other β-lactams as well.⁸¹ Elevated serum levels of semisynthetic penicillins or cephalosporins may cause hemostatic defects in patients with impaired renal failure because of interference with platelet function.⁸²^,⁸³ Elevated serum levels of aminoglycosides may result in eighth nerve damage.³⁸^,⁸⁰ Neurotoxic reactions, including respiratory arrest and death, may occur in patients who have excessive serum levels of certain aminoglycosides or the polymyxins.³⁴^,⁸⁴ Bone marrow suppression may occur in patients with renal failure who receive inappropriately high doses of 5-fluorocytosine.⁸⁵ In all these situations and numerous others, the possibility of toxic reactions can be lessened significantly or eliminated if the doses of the antimicrobial agents are appropriately reduced in the presence of renal insufficiency.

The tetracyclines, except doxycycline, tigecycline, and possibly minocycline, are contraindicated in patients with impaired renal function because the elevated serum levels that result may produce a significant worsening of the uremic state resulting from their antianabolic effect. Moreover, they may cause enhanced hepatotoxicity in this situation.³⁴ The long-acting sulfonamides should be avoided in this situation because they are potentially nephrotoxic.

Certain antimicrobial agents, including erythromycin, azithromycin, chloramphenicol, lincomycin, and clindamycin, should be used with caution in patients with impaired hepatic function.⁸⁶ These drugs are primarily excreted or detoxified in the liver. Bone marrow suppression caused by chloramphenicol is much more likely to occur in patients with impaired hepatic function; because of this, it has been suggested that the dose of chloramphenicol be cut at least in half in patients with cirrhosis and other severe liver disease.⁸⁷ Because the serum half-life of clindamycin is increased in patients with severe liver disease, its dose should also be decreased in this situation. The tetracyclines may produce elevations in serum transaminase levels in patients recovering from viral hepatitis.³⁴ They should be avoided or used with extreme caution in patients with underlying liver disease. The serum half-lives of both rifampin and isoniazid are prolonged in patients with cirrhosis.⁸⁸ Other drugs that should be used with caution, or for which serum levels should be monitored and/or doses adjusted in patients with severe liver disease, include metronidazole, fluconazole, itraconazole, nitrofurantoin, and pyrazinamide.⁸⁶ Dose adjustments are also required for newer agents, such as tigecycline and voriconazole, in patients with advanced liver disease. It has been suggested that β-lactam antibiotic–induced leukopenia occurs more frequently in patients with impaired hepatic function.⁸⁹ Hepatobiliary disease influences antimicrobial therapy in still another way. The biliary concentrations of many antimicrobial agents, including ampicillin and nafcillin, that are normally excreted in high concentration in the bile may be reduced significantly in patients with liver disease or biliary obstruction.³⁴

Site of Infection

Another consideration in selection of an appropriate antimicrobial is the site of infection. For antimicrobial therapy to be effective, an adequate concentration of the drug must be delivered to the site of infection. In most cases, this means that the local concentration of the antimicrobial agent should at least equal the MIC of the infecting organism. Concentrations representing multiples of the MIC are generally believed more likely to be efficacious, but in many cases, such local concentrations may be difficult or impossible to achieve. A failure to achieve local concentrations of antibiotics higher than the MIC of the infecting organism may not always result in failure, however, because there is evidence that subinhibitory concentrations of drugs may produce antimicrobial effects that aid the host defenses against infections. It has been demonstrated clearly that subinhibitory concentrations of antibiotics can alter bacterial morphology,⁹⁰ adherence properties,⁹¹ and opsonic requirements⁹²; can enhance phagocytosis⁹³; and can even aid intracellular killing of bacteria by polymorphonuclear leukocytes.⁹⁴ This may explain the clinical observation that, on occasion, doses of antimicrobials that produce seemingly inadequate serum levels may still result in clinical cure. In spite of such observations, most infectious disease clinicians feel that optimal therapy requires concentrations of antimicrobials that are above the MIC at the site of infection (see Chapter 19).

Serum concentrations of antimicrobial agents are, in principle, relatively easy to determine. Nevertheless, monitoring of serum concentrations is routinely performed only for a limited number of antimicrobials, such as vancomycin, aminoglycosides, and 5-flucytosine. With the latter agents, monitoring is performed primarily to help minimize the risk of toxicity from overly high concentrations. Recent data suggest that therapeutic drug monitoring may have an increasing role in the management of fungal infections.⁹⁵ However, except in cases of bacteremia, antimicrobial efficacy is more likely determined by the tissue concentration than by the blood level, as noted earlier. Some agents such as spiramycin, certain macrolides such as azithromycin, and tigecycline can be effective in some infections, despite an inability to achieve serum levels above the MIC of certain organisms.⁹⁶^,⁹⁷ This may be explained by their ability to achieve intracellular and tissue concentrations that far exceed those obtained in serum.⁹⁸^,⁹⁹ Binding to serum proteins may affect both the tissue distribution and the activity of antimicrobial agents in the blood.

Although much careful investigation has been done on protein binding, the precise clinical significance of this phenomenon remains to be determined. For example, it has been shown that only the unbound form of a given antimicrobial agent is active in vitro (and presumably also in vivo) against infecting organisms.¹⁰⁰ However, because protein binding can be rapidly reversible,¹⁰¹ the activity of even highly protein-bound agents may not be limited absolutely by protein binding. The penetration of antimicrobial agents into interstitial fluid and lymph is related to protein binding because only the free form of the agent is able to pass through the capillary wall.¹⁰⁰ Penetration of antibiotics into fibrin clots, which may be analogous to the penetration of the drugs to reach the site of infection in patients with bacterial endocarditis, is likewise related to the amount of unbound antibiotic in the surrounding fluid.¹⁰² Nevertheless, it is often difficult to correlate therapeutic outcome with in vitro susceptibility (MIC) and protein binding alone.¹⁰³^,¹⁰⁴ Several technical factors also contribute to problems with such correlations because the protein binding measured in vitro may vary with the concentration of antibiotic tested and with other variables, such as medium composition, pH, and temperature.¹⁰⁵

In recent years, there has been growing appreciation that the profile of antimicrobial concentrations over time relative to susceptibility of the pathogen is critically important to the effectiveness of antimicrobial therapy. For example, the effectiveness of β-lactam antibiotics can be best correlated with attainment of unbound (free) drug concentrations greater than the MIC of the pathogen for a certain proportion of the dosing interval; that is, the effectiveness is dependent on time interval of free-drug concentrations greater than the MIC. In contrast, the effectiveness of fluoroquinolones correlates better with the ratio of the free-drug “area under the concentration-time curve” (AUC) to the MIC of the organism. Thus, the effectiveness of an antimicrobial agent depends on a number of factors, including organism susceptibility, drug class (to establish time-dependent or concentration-dependent behavior), and protein binding.

The ability of an antibiotic to penetrate to the site of infection with an appropriate pharmacodynamic profile is a major determinant in the successful therapy. The ability of an antibiotic to pass through membranes by nonionic diffusion is related to its lipid solubility. Thus, lipid-soluble agents, such as chloramphenicol, rifampin, trimethoprim, and isoniazid, are all more efficient in penetrating membranes than are the more highly ionized compounds.¹⁰⁰ These agents rapidly cross the blood-brain barrier and produce better cerebrospinal fluid levels than do more highly ionized compounds, such as the aminoglycosides.

Agents that are excreted by the liver and are concentrated in the bile, such as ampicillin and doxycycline, may be more effective in treating cholangitis than are agents such as the first-generation cephalosporins or aminoglycosides, which are not greatly concentrated in bile. The new fluoroquinolones may owe some of their effectiveness in the treatment of osteomyelitis to their ability to achieve superior concentrations in bone.¹⁰⁶ These agents penetrate far more effectively into the prostate than the β-lactams or aminoglycosides, and this undoubtedly contributes to their superior therapeutic efficacy in prostatitis.¹⁰⁷^,¹⁰⁸

Even the achievement of “therapeutic concentrations” of antimicrobial agents at the site of infection may not be sufficient for cure because a number of local factors may influence the activity of antimicrobial agents. An important example is the binding and inactivation of daptomycin by pulmonary surfactant. Although this drug penetrates well into pulmonary epithelial lining fluid, and is highly active against pneumococci in vitro, clinical trials for treatment of community-acquired pneumonia demonstrated daptomycin to be less effective than standard treatment with ceftriaxone.¹⁰⁹ The reason for this was subsequently shown to be that daptomycin is bound by pulmonary surfactant and, in this bound state, demonstrates poor antimicrobial activity.¹¹⁰ Aminoglycosides and the polymyxins are bound to and inactivated by purulent material.¹¹¹ This is one of many reasons why surgical drainage is imperative when treating abscesses with agents such as these. Although penicillins may be more active in purulent material, clinical experience strongly suggests that appropriate drainage procedures greatly enhance the efficacy of these agents as well. Penicillin G is not inactivated by purulent material per se, but the presence of β-lactamase–producing organisms, such as Bacteroides fragilis, in abscesses may result in local inactivation of penicillin G and other β-lactam antibiotics.¹¹² Penicillins and tetracyclines are also bound by hemoglobin and thus may be less effective in the presence of significant hematoma formation.¹⁰⁰

Local decreases in oxygen tension, such as occur in abscesses and intraperitoneal infections, may also have an effect on the activity of certain antimicrobial agents. The aminoglycosides, for example, are inactive against anaerobes and may also be less effective against facultative organisms under anaerobic conditions because oxygen is required for the transport of these agents into the bacterial cell.¹¹³

Local alterations in pH, such as occur in abscesses and especially in the urine, may have an important effect on the activity of a number of antimicrobial agents. Methenamine and nitrofurantoin are more active at an acid pH, whereas alkalinization enhances the activity of erythromycin, azithromycin, clarithromycin, lincomycin, clindamycin, and the aminoglycosides. Indeed, the aminoglycosides show a marked loss of activity at a low pH. These observations have occasionally been used to advantage in treating patients with urinary tract infections, a situation in which the local pH can be altered by the addition of acidifying or alkalinizing agents.¹¹⁴^,¹¹⁵

The presence of foreign bodies also has a profound effect on the activity of antimicrobial agents. Thus, it is sometimes necessary to remove foreign material to cure an infection of a prosthetic heart valve and almost always necessary to remove, or at least to débride carefully, prosthetic devices for cure of joint implant infections.¹¹⁶ The mechanism by which foreign bodies potentiate infection is not clear, but they probably cause localized impairment of host defense mechanisms.¹¹⁷ In addition, the foreign body often serves as a nidus on which organisms can adhere and produce extracellular substances, such as glycocalyx or biofilm, which may interfere with phagocytosis.¹¹⁸ Although it was originally thought that biofilm produces a barrier to penetration of antimicrobials, this is clearly not the case. The ineffectiveness of antibiotics against bacteria in biofilm is the result of alterations in the metabolic state of these organisms that renders them relatively resistant to the action of antibiotics.¹¹⁸^–¹²⁰

Antimicrobial agents themselves have the potential to alter host defenses. Clinically achievable concentrations of many different agents have been shown to diminish leukocyte chemotaxis, lymphocyte transformation, monocyte transformation, delayed hypersensitivity, antibody production, phagocytosis, and the microbicidal action of polymorphonuclear leukocytes.¹²¹^–¹²⁷ It is not clear, however, whether any of these effects (largely demonstrated by in vitro studies) are of clinical significance.¹²⁷ In fact, the modification of biologic responses by certain antibiotics may have beneficial effects as well.¹²⁸^,¹²⁹ Nonetheless, the possibility that antimicrobial agents can cause immunosuppression exists, and this fact should discourage the indiscriminate use of antibiotics, especially in patients who are already immunosuppressed because of their underlying disease or concomitant drug therapy.¹²⁵ Finally, antimicrobial agents, such as the β-lactams, that cause rapid lysis of bacteria may also release endotoxins or cell wall components that have potentially deleterious local or systemic effects, or both, in the host. The local inflammatory consequences of such activity have been clearly defined in experimental models of bacterial meningitis (which forms the basis for the use of dexamethasone in bacterial meningitis),¹³⁰^,¹³¹ but their significance in other settings, such as gram-negative sepsis, remains to be determined.¹³² The use of antibiotics early in the course of intestinal infection with E. coli O157:H7 was found to be a risk factor for the subsequent development of hemolytic-uremic syndrome, consistent with findings that antibiotics may stimulate production of Shiga toxin from these organisms in vitro.¹³³