GENERAL FEATURES AND PRINCIPLES
The post-antibiotic era has begun. For most people, this is the first time in their lives that an effective treatment for a bacterial infection may not exist. The Enterobacteriaceae are at the forefront of this evolving public health crisis. For example, the Centers for Disease Control and Prevention (CDC) and the World Health Organization (WHO) have designated carbapenem-resistant Enterobacteriaceae as representing a threat level of “urgent” and “priority one, critical,” respectively. Enterobacteriaceae are responsible for a significant proportion of the deaths attributed to resistant bacteria, the number of which has been estimated at 23,000 and 25,000 annually in the United States and the European Union, respectively, with numbers three- to fivefold greater (per capita) in low- and middle-income countries (e.g., Thailand). These pathogens cause a wide variety of infections involving diverse anatomic sites in both healthy and compromised hosts. Therefore, a thorough knowledge of clinical presentations and appropriate therapeutic choices is necessary for optimal outcomes. Escherichia coli, Klebsiella, Proteus, Enterobacter, Serratia, Citrobacter, Morganella, Providencia, Cronobacter, and Edwardsiella are enteric gram-negative bacilli (GNB) that are members of the family Enterobacteriaceae. Salmonella, Shigella, and Yersinia, also in the family Enterobacteriaceae, are discussed in Chaps. 160, 161, and 166, respectively.
E. coli, Klebsiella, Proteus, Enterobacter, Serratia, Citrobacter, Morganella, Providencia, Cronobacter, and Edwardsiella are components of the normal animal and human colonic microbiota and/or the microbiota in various environmental habitats, including long-term-care facilities (LTCFs) and hospitals. As a result, except for certain pathotypes of intestinal pathogenic E. coli, these genera are global pathogens. The incidence of infection due to these agents is increasing because of the combination of an aging population and increasing antimicrobial resistance. In healthy humans, E. coli is the predominant species of GNB in the colonic microbiota; Klebsiella and Proteus are less prevalent. GNB (primarily E. coli, Klebsiella, and Proteus) colonize the oropharynx and skin of healthy individuals only transiently. By contrast, in LTCFs and hospital settings, a variety of GNB emerge as the dominant colonizers of both mucosal and skin surfaces, particularly in association with antimicrobial use, severe illness, and extended length of stay. LTCFs are emerging as an important reservoir for resistant GNB. This colonization may lead to subsequent infection; for example, oropharyngeal colonization may lead to pneumonia, and colonic/perineal colonization may lead to urinary tract infection (UTI). The use of ampicillin or amoxicillin was associated with an increased risk of subsequent infection due to the hypervirulent pathotype of Klebsiella pneumoniae in Taiwan; this association suggests that changes in the quantity or prevalence of colonizing bacteria may significantly influence the risk of infection. Serratia and Enterobacter infection may be acquired directly through a variety of infusates (e.g., medications, blood products). Edwardsiella infections are acquired through freshwater and marine environment exposures and are most common in Southeast Asia.
Enteric GNB possess an extracytoplasmic outer membrane consisting of a lipid bilayer with associated proteins, lipoproteins, and polysaccharides (capsule, lipopolysaccharide). The outer membrane interfaces with the external environment, including the human host. A variety of components of the outer membrane are critical determinants in pathogenesis (e.g., capsule) and antimicrobial resistance (e.g., permeability barrier, efflux pumps). In addition, secreted products play an important role in both host infection (e.g., iron acquisition molecules) and environmental niche survival and colonization (e.g., type VI secretion systems).
Multiple bacterial virulence factors are required for the pathogenesis of infections caused by GNB. Possession of specialized virulence genes defines pathogens and enables them to infect the host efficiently. Hosts and their cognate pathogens have been co-adapting throughout evolutionary history. During the host–pathogen “chess match” over time, various and redundant strategies have emerged in both the pathogens and their hosts (Table 156-1).
TABLE 156-1Interactions of Extraintestinal Pathogenic Escherichia coli with the Human Host: A Paradigm for Extracellular, Extraintestinal Gram-Negative Bacterial Pathogens ||Download (.pdf) TABLE 156-1 Interactions of Extraintestinal Pathogenic Escherichia coli with the Human Host: A Paradigm for Extracellular, Extraintestinal Gram-Negative Bacterial Pathogens
|Bacterial Goal ||Host Obstacle ||Bacterial Solution |
|Extraintestinal attachment ||Flow of urine, mucociliary escalator ||Multiple adhesins (e.g., type 1, S, and F1C fimbriae; P pili) |
|Nutrient acquisition for growth ||Nutrient sequestration (e.g., iron via intracellular storage and extracellular scavenging via lactoferrin and transferrin) ||Cellular lysis (e.g., hemolysin), multiple mechanisms for competing for iron (e.g., siderophores) and other nutrients |
|Initial avoidance of host bactericidal activity ||Complement, phagocytic cells, antimicrobial peptides ||Capsular polysaccharide, lipopolysaccharide |
|Dissemination (within host and between hosts) ||Intact tissue barriers ||Irritant tissue damage resulting in increased excretion (e.g., toxins such as hemolysin), invasion of brain endothelium |
|Late avoidance of host bactericidal activity ||Acquired immunity (e.g., specific antibodies), treatment with antibiotics ||Cell entry, acquisition of antimicrobial resistance |
Intestinal pathogenic (diarrheagenic) mechanisms are discussed below. The members of the Enterobacteriaceae family that cause extraintestinal infections are primarily extracellular pathogens and therefore share certain pathogenic features. The principal components of host defense against Enterobacteriaceae, regardless of species, are innate immunity (including intact skin and mucosal barriers; the withholding of nutrients; and the activities of complement, antimicrobial peptides, and professional phagocytes) and humoral immunity. Both susceptibility to and severity of infection are increased with dysfunction or deficiencies of these host components. By contrast, the virulence traits of intestinal pathogenic E. coli—i.e., the distinctive strains that can cause diarrheal disease—are for the most part different from those of extraintestinal pathogenic E. coli (ExPEC) and other GNB that cause extraintestinal infections. This distinction reflects site-specific differences in host environments and defense mechanisms.
A given enterobacterial strain usually possesses multiple adhesins for binding to a variety of host cells (e.g., in E. coli: type 1, S, and F1C fimbriae; P pili). Nutrient acquisition (e.g., of iron via siderophores) requires many genes that are necessary but not sufficient for pathogenesis. The ability to resist the bactericidal activity of complement and phagocytes in the absence of antibody (e.g., as conferred by capsule or the O antigen component of lipopolysaccharide) is one of the defining traits of an extracellular pathogen. Tissue damage (e.g., as mediated by E. coli hemolysin) may facilitate nutrient acquisition and spread within the host. Without doubt, many important virulence genes await identification (Chap. 116).
The ability to induce septic shock is another defining feature of these genera. GNB are the most common causes of this potentially lethal syndrome. Pathogen-associated molecular pattern molecules (PAMPs; e.g., the lipid A moiety of lipopolysaccharide) stimulate a proinflammatory host response via pattern recognition receptors (e.g., Toll-like or C-type lectin receptors) that activate host defense signaling pathways; if overly exuberant, this response results in shock (Chap. 297). Direct bacterial damage of host tissue (e.g., by toxins) or collateral damage from the host response can result in the release of damage-associated molecular pattern molecules (DAMPs; e.g., HMGB1) that can propagate a detrimental proinflammatory host response.
Many antigenic variants (serotypes) exist in most genera of GNB. For example, E. coli has more than 150 O (somatic) antigens, 80 K (capsular) antigens, and 53 H (flagellar) antigens. This antigenic variability, which permits immune evasion and allows recurrent infection by different strains of the same species, has impeded vaccine development (Chap. 118).
Depending on both the host and the pathogen, GNB can infect nearly every organ or body cavity. E. coli can cause either intestinal or extraintestinal infection, depending on the particular pathotype, and Edwardsiella tarda can cause both intestinal and extraintestinal infection. Klebsiella causes primarily extraintestinal infection, but a toxin-producing variant of Klebsiella oxytoca has been associated with hemorrhagic colitis.
E. coli and—to a lesser degree—Klebsiella account for most extraintestinal infections due to GNB. These species (for K. pneumoniae, primarily its hypervirulent pathotype) are the most virulent pathogens within this group, as demonstrated by their ability to cause severe infections in healthy, ambulatory hosts from the community. However, the other genera of GNB are also important extraintestinal pathogens, especially among LTCF residents and hospitalized patients, in large part because of the intrinsic or acquired antimicrobial resistance of these organisms and the increasing number of individuals with compromised host defenses. The mortality rate is substantial in many GNB infections and correlates with severity of illness and underlying host status. Especially problematic are pneumonia, sepsis, and septic shock (arising from any site of infection), for which the associated mortality rates are 20–60%.
Isolation of GNB from sterile sites almost always implies infection, whereas their isolation from nonsterile sites, particularly from open wounds and the respiratory tract, requires clinical correlation to differentiate colonization from infection. Clinical microbiology laboratories are increasingly incorporating newer molecular-based methodologies (e.g., matrix-assisted laser desorption–ionization–time-of-flight mass spectrometry [MALDI-TOF-MS] and polymerase chain reaction [PCR]) to enhance the sensitivity, accuracy, and rapidity of reporting on pathogen identification and resistance genes (e.g., blaKPC, NDM, OXA, CTX). This information can be used to increase the timeliness of initiation and/or the accurate selection of empirical antimicrobial therapy, thereby improving outcomes.
TREATMENT Infections Caused by Gram-Negative Enteric Bacilli
Initiation of appropriate empirical antimicrobial therapy early in the course of GNB infections (particularly serious ones) leads to improved outcomes (See also Chap. 139). The ever-increasing prevalence of multidrug-resistant (MDR) and extensively drug-resistant (XDR) GNB; the lag between published and current resistance rates; and variations in antimicrobial susceptibility by species, geographic location, regional antimicrobial use, and hospital site (e.g., intensive care units [ICUs] versus wards) necessitate familiarity with evolving patterns of antimicrobial resistance for the selection of appropriate empirical therapy. Factors predictive of resistance in a given isolate include recent antimicrobial use, a health care association (e.g., recent or ongoing hospitalization, dialysis, residence in an LTCF), or international travel (e.g., to Asia, Latin America, Africa, Eastern Europe). Resistance rates will almost certainly increase over time and will likely be higher than shown here by the time this chapter is published. Data for 2008–2014 from the U.S. National Healthcare Safety Network indicates that the prevalence of the extended-spectrum β-lactamase (ESBL) phenotype among Enterobacteriaceae isolates varied by health care setting—i.e., 16% for short-term care, 38.6% for long-term care, and 10.7% for inpatient rehabilitation facilities—as did the prevalence of carbapenem resistance (2.8%, 12%, and 1.9%, respectively). Global ESBL rates for Enterobacteriaceae isolates from hospitalized patients were roughly similar in North America, Western Europe, Australia, and New Zealand and were higher in Latin America, Eastern Europe, and Asia. Perhaps even more concerning is the reported isolation of carbapenem-resistant Enterobacteriaceae (mediated primarily by New Delhi metallo-β-lactamase [NDM]) from ambulatory patients without known risk factors.
For appropriately selected patients, it may be prudent initially, pending antimicrobial susceptibility results, to use two potentially active agents as a way to increase the likelihood that at least one agent will be active against the patient’s organism. If broad-spectrum treatment has been initiated, it is important to switch to the most appropriate narrower-spectrum agent once antimicrobial susceptibility results become available. Such responsible antimicrobial stewardship should help disrupt the ever-escalating cycle of selection for increasingly resistant bacteria, decrease the likelihood of Clostridium difficile infection, decrease costs, and maximize the useful longevity of available antimicrobial agents. Likewise, it is important to avoid treatment of patients who are colonized but not infected (e.g., who have a positive sputum culture without evidence of pneumonia or a positive urine culture without clinical manifestations of UTI).
At present, the most reliably active antimicrobial agents against GNB are the carbapenems (e.g., meropenem); the aminoglycoside amikacin; the fourth-generation cephalosporin cefepime; the β-lactamase inhibitor combination agents piperacillin-tazobactam, ceftolozane-tazobactam, and ceftazidime-avibactam; and the polymyxins (colistin and polymyxin B). However, it should be noted that Proteus, Serratia, Morganella, and Providencia are intrinsically resistant to the polymyxins. The number of antimicrobial agents effective against certain Enterobacteriaceae is shrinking, and truly pan-resistant GNB exist. Accordingly, the currently available antimicrobial drugs must be used judiciously.
β-Lactamases, which inactivate β-lactam agents, are the most important mediators of β-lactam resistance in GNB. Decreased permeability and/or active efflux of β-lactam agents, although less important and less potent, may occur alone or in combination with β-lactamase-mediated resistance.
Broad-spectrum β-lactamases (e.g., TEM, SHV), which mediate resistance to many penicillins and first-generation cephalosporins, are frequently expressed in enteric GNB. These enzymes are inhibited by β-lactamase inhibitors (e.g., clavulanate, sulbactam, tazobactam, avibactam). In their wild-type form, they do not hydrolyze third- and fourth-generation cephalosporins or cephamycins (e.g., cefoxitin). However, molecular variants of TEM and SHV that have amino acid replacements at certain critical positions in the peptide do exhibit such hydrolytic capability and thus are referred to as ESBLs, as discussed below.
ESBLs (e.g., CTX-M, SHV, TEM) are modified broad-spectrum enzymes that hydrolyze third-generation cephalosporins, aztreonam, and (in some instances) fourth-generation cephalosporins in addition to the drugs hydrolyzed by broad-spectrum β-lactamases. GNB that express ESBLs may also possess porin mutations that result in decreased uptake of cephalosporins, β-lactam/β-lactamase inhibitor combinations, and carbapenems, thereby further reducing susceptibility to these β-lactam agents. The prevalence of acquired ESBL production, particularly of CTX-M-type enzymes, is increasing in GNB worldwide, in large part as a result of the presence of the corresponding genes on transferable (conjugal) plasmids variably linked to or associated with resistance to fluoroquinolones, trimethoprim-sulfamethoxazole (TMP-SMX), aminoglycosides, tetracyclines, and (more recently) fosfomycin.
To date, ESBLs are most prevalent in E. coli, K. pneumoniae, and K. oxytoca, but these enzymes can occur in all Enterobacteriaceae. The approximate regional prevalence of ESBL-producing GNB currently follows a descending gradient as follows: China > Eastern Europe > other parts of Asia > Latin America and Africa > Western Europe, the United States, Canada, and Australia. International travel to high-prevalence regions increases the likelihood of colonization with these strains.
ESBL-producing GNB were described initially in hospitals (ICUs > wards) and LTCFs, where outbreaks occurred in association with extensive use of third-generation cephalosporins. However, over the last decade, the incidence of UTI due to CTX-M ESBL-producing E. coli has increased worldwide (including in the United States), even among healthy ambulatory women without health care or antimicrobial exposure. Antimicrobial use in food animals has also been implicated in the rise of ESBLs.
Carbapenems are the most reliably active β-lactam agents against ESBL-expressing strains. Clinical experience with alternatives is more limited, but, for organisms susceptible to piperacillin-tazobactam (minimal inhibitory concentration [MIC], ≤4 μg/mL), this agent—dosed at 4.5 g q6h—may offer a carbapenem-sparing alternative, as may ceftazidime-avibactam and ceftolozane-tazobactam.
The role of tigecycline is unclear despite its excellent in vitro activity; Proteus, Morganella, and Providencia are inherently resistant, and attainable serum and urine levels are low. Therefore, caution is advisable, especially with serious infections, until more clinical data become available.
Oral options for the treatment of strains expressing ESBLs are limited. Fosfomycin and nitrofurantoin (for E. coli) and perhaps pivmecillinam (not available in the United States) are the most reliably active agents.
AmpC β-lactamases, when induced or stably derepressed to high levels of expression, confer resistance to the same substrates as do ESBLs as well as to the cephamycins (e.g., cefoxitin and cefotetan). The genes encoding these enzymes are primarily chromosomal and therefore may not exhibit the linked or associated resistance to fluoroquinolones, TMP-SMX, aminoglycosides, and tetracyclines that is common with ESBLs. These enzymes are problematic for the clinician: resistance may develop during therapy with third-generation cephalosporins and result in clinical failure, particularly in the setting of bacteremia.
Although chromosomal AmpC β-lactamases are present in nearly all members of the Enterobacteriaceae family, the risk of clinically significant induction of high-level expression or selection of stably derepressed mutants with cephalosporin treatment is greatest with Enterobacter cloacae and Enterobacter aerogenes, lower with Serratia marcescens and Citrobacter freundii, and lowest with Providencia and Morganella morganii. In addition, rare strains of E. coli, K. pneumoniae, and other Enterobacteriaceae have acquired plasmids containing inducible AmpC β-lactamase genes.
For AmpC-expressing strains, carbapenems are an appropriate treatment option. Ceftazidime-avibactam and ceftolozane-tazobactam are active in vitro, but clinical data are limited. The fourth-generation cephalosporin cefepime may be an appropriate option if the concomitant presence of an ESBL can be excluded (a task that currently exceeds the capability of most clinical microbiology laboratories) and source control is achieved. Although clinical data are limited, other carbapenem-sparing alternatives to consider if isolates are susceptible in vitro include fluoroquinolones, piperacillin-tazobactam, TMP-SMX, tigecycline, and aminoglycosides.
Carbapenemases—e.g., class A (Klebsiella pneumoniae carbapenemase [KPC]); class B (NDM; Verona integron–encoded metallo-β-lactamase [VIM]; and imipenemase metallo-β-lactamase [IMP]); and class D [OXA-48])—confer resistance to the same drugs as do ESBLs as well as to cephamycins and carbapenems. As with ESBLs, carbapenemase-encoding genes may be present on transferable plasmids, which often encode linked resistance to fluoroquinolones, TMP-SMX, tetracyclines, and aminoglycosides. Transposon-mediated spread (e.g., TN4401 for KPC) is also important. Unfortunately, carbapenemase-producing Enterobacteriaceae are becoming increasingly common, particularly in Asia. Asymptomatic intestinal carriage may facilitate spread.
Carbapenemase production by Enterobacteriaceae is most prevalent in K. pneumoniae and, secondarily, in E. coli, but has been described in nearly all members of the family. Carbapenem resistance may also occur in the absence of carbapenemase production, mediated by production of an AmpC β-lactamase and/or ESBL along with modifications in permeability/efflux. Resistance to any carbapenem should prompt assessment for carbapenemase production via either genotypic or phenotypic tests, if available; the exception to this rule is isolated resistance to imipenem in M. morganii, Proteus, and Providencia, which exhibit intrinsic low-level resistance. Although the modified Hodge test is used widely for phenotypic confirmation of carbapenemase production, its limitations include false-positive results with Enterobacter species and false-negative results with NDM.
For treatment of infections due to carbapenem-resistant Enterobacteriaceae, tigecycline and colistin are the most reliably active parenteral agents in vitro. However, because tigecycline reaches only low serum and urine concentrations, caution is warranted in using it to treat bacteremia and perhaps UTI, although a few case reports describe some success with tigecycline therapy for UTI. Colistin has nephrotoxic and neurotoxic potential. The recent emergence of the colistin resistance gene mcr-1 on a stable transferable plasmid is extremely concerning since polymyxins (polymyxin B and polymyxin E [colistin]) currently constitute a last line of defense against strains that produce metallo-carbapenemases (e.g., NDM-1). In addition, in a recent study, 13% of carbapenem-resistant K. pneumoniae isolates were co-resistant to colistin independent of mcr-1.
Ceftazidime-avibactam is active in vitro against the serine carbapenemases (e.g., KPC, OXA-48) but not the metallo-carbapenemases (e.g., NDM, VIM, IMP). However, limited clinical data from an uncontrolled retrospective study of ceftazidime-avibactam for the treatment of infection with carbapenem-resistant Enterobacteriaceae demonstrated suboptimal efficacy and development of resistance in 8% of the cohort. Aminoglycosides may have some utility for combination therapy if they are active in vitro. Fosfomycin is often active in vitro, but clinical data are limited. Furthermore, resistance may develop with monotherapy and increased use, plasmid-mediated resistance (via fosA3) has been described (raising concern about rapid dissemination), resistance is generally more prevalent among XDR strains, susceptibility testing may not be readily available, and no parenteral formulation is available in the United States. Aztreonam is active against the problematic metallo-carbapenemases but is hydrolyzed by ESBLs and AmpC β-lactamases, which often co-exist in XDR strains. Ongoing studies are assessing aztreonam plus avibactam, a promising combination for the treatment of pan-drug-resistant strains.
Extensive resistance to available agents may leave the clinician with few or no ideal therapeutic options. However, use of a regimen that takes into account the site of infection, achievable drug levels at that site (e.g., higher concentrations of many agents in urine), and pharmacodynamic factors (e.g., prolonged infusion of β-lactam agents to maintain drug levels above the MIC) may increase the chance for a successful outcome. Likewise, observational data suggest that combination therapy may be beneficial against carbapenem-resistant Enterobacteriaceae; randomized controlled trials are in progress.
Resistance to fluoroquinolones usually is due to alterations in or protection of the target sites in DNA gyrase and topoisomerase IV, with or without decreased permeability and active efflux. Fluoroquinolone resistance is increasingly prevalent among GNB and is associated with resistance to other antimicrobial classes; for example, 20–80% of ESBL-producing enteric GNB are also resistant to fluoroquinolones. At present, fluoroquinolones should be considered unreliable as empirical therapy for GNB infections in critically ill patients.
In this era of increasing antimicrobial resistance, it is critical to culture the primary site of infection before initiating antimicrobial therapy and, for systemically ill patients, to obtain blood cultures. In vitro testing may not always detect antimicrobial resistance; therefore, it is important to assess the patient’s clinical response to treatment. Moreover, as discussed above, resistance may emerge during therapy through the induction or stable derepression of AmpC β-lactamases. In addition, drainage of abscesses, resection of necrotic tissue, and removal of infected foreign bodies, sometimes referred to collectively as “source control,” are often required for cure.
GNB are commonly involved in polymicrobial infections, in which the role of each individual pathogen is uncertain (Chap. 172). Although some GNB are more pathogenic than others, it is usually prudent, if possible, to design an antimicrobial regimen active against all of the GNB identified, because each is capable of pathogenicity in its own right. Lastly, for patients treated initially with a broad-spectrum empirical regimen, the regimen should be de-escalated as expeditiously as possible once susceptibility results are known and the patient has responded to therapy.
Certain measures are broadly applicable for decreasing infection risk (See also Chap. 137). Antimicrobial stewardship programs should be instituted to facilitate appropriate antimicrobial use, which will minimize the development of resistance. Diligent adherence to hand-hygiene protocols by health care personnel and cleaning/disinfection or single-patient use of objects that come into contact with patients (e.g., stethoscopes and blood pressure cuffs) are essential. Indwelling devices (e.g., urinary and intravascular catheters) should be used only when necessary and inserted according to an appropriate protocol; protocols for daily-use evaluation and prompt removal should be implemented. Multi-use medication vials should be avoided if possible. Oral application of chlorhexidine decreases the incidence of pneumonia among patients on ventilators. Increasing data support the implementation of universal decolonization (e.g., chlorhexidine bathing) to prevent infection in ICU patients. The public health threat from carbapenem-resistant Enterobacteriaceae has resulted in additional recommendations, especially for carbapenemase-producing carbapenem-resistant Enterobacteriaceae, which are an even greater concern. These recommendations include contact precautions for patients colonized or infected with carbapenem-resistant Enterobacteriaceae, notification to the receiving facility from facilities transferring a patient colonized or infected with these organisms, and daily environmental cleaning. Screening of contacts and active surveillance for these bacteria may also be appropriate.
ESCHERICHIA COLI INFECTIONS
All E. coli strains share a core genome of ~2000 genes. In contrast, an E. coli strain’s ability to cause infection and the nature of such infections are defined largely by accessory (i.e., non-core, non-essential) genes that encode various virulence factors. The composition of the E. coli accessory genome is fluid and ongoing, as demonstrated by the recent evolution of Shiga toxin–producing enteroaggregative E. coli.
Commensal E. coli variants are an important constituent of the normal intestinal microbiota that confer benefits to the host (e.g., resistance to colonization with pathogenic organisms). Such strains generally lack the specialized virulence traits that enable extraintestinal and intestinal pathogenic E. coli strains to cause disease outside and within the gastrointestinal tract, respectively. However, even commensal E. coli strains can be involved in extraintestinal infections in the presence of an aggravating factor, such as a foreign body (e.g., a urinary catheter), host compromise (e.g., local anatomic or functional abnormalities [including urinary or biliary tract obstruction] or systemic immunocompromise), or an inoculum that is large or contains a mixture of bacterial species (e.g., fecal contamination of the peritoneal cavity).
EXTRAINTESTINAL PATHOGENIC STRAINS
ExPEC strains are the most common enteric GNB to cause community-acquired and health care–associated bacterial infections. The emerging propensity of these strains to acquire new mechanisms of antimicrobial resistance (e.g., ESBLs and carbapenemases) poses novel challenges in managing ExPEC infection. Several ExPEC clonal groups (e.g., ST131, ST95, ST69, and ST73) are recognized to have undergone global dissemination. The mechanisms underlying the epidemiologic success of such disseminated lineages presumably include superior biological fitness and acquisition of antimicrobial resistance, as demonstrated by members of the ST131 subclone H30-Rx, which are resistant to fluoroquinolones and usually express the ESBL CTX-M-15. Reservoirs and transmission pathways are an active area of study, but human-to-human, food-to-human (e.g., pork, turkey, and chicken), and perhaps environment-to-human are most likely.
Like commensal E. coli (but unlike intestinal pathogenic E. coli), ExPEC strains are often found in the intestinal microbiota of healthy individuals and do not cause gastroenteritis in humans. Entry from their site of colonization (e.g., the colon, vagina, or oropharynx) into a normally sterile extraintestinal site (e.g., the urinary tract, peritoneal cavity, or lungs) is the rate-limiting step for infection. ExPEC strains have acquired accessory genes encoding diverse extraintestinal virulence factors that enable the bacteria to cause infections outside the gastrointestinal tract in both normal and compromised hosts (Table 156-1). These virulence genes define ExPEC and, for the most part, are distinct from the virulence genes that enable intestinal pathogenic strains to cause diarrheal disease (Table 156-2). All age groups, all types of hosts, and nearly all organs and anatomic sites are susceptible to infection by ExPEC. Even previously healthy hosts can become severely ill or die when infected with ExPEC; however, adverse outcomes are more common among hosts with comorbid illnesses and host defense abnormalities. The diversity and the medical and economic impact of ExPEC infections are evident from consideration of the following specific syndromes.
TABLE 156-2Intestinal Pathogenic E. coli ||Download (.pdf) TABLE 156-2 Intestinal Pathogenic E. coli
|Pathotype ||Epidemiology ||Clinical Syndromea ||Defining Molecular Trait ||Responsible Genetic Elementb |
|STEC/EHEC/ST-EAEC ||Food, water, person-to-person; all ages, industrialized countries ||Hemorrhagic colitis, hemolytic-uremic syndrome ||Shiga toxin ||Lambda-like Stx1- or Stx2-encoding bacteriophage |
|ETEC ||Food, water; young children in and travelers to developing countries ||Traveler’s diarrhea ||Heat-stable and labile enterotoxins, colonization factors ||Virulence plasmid(s) |
|EPEC ||Person-to-person; young children and neonates in developing countries ||Watery diarrhea, persistent diarrhea ||Localized adherence, attaching and effacing lesion on intestinal epithelium ||EPEC adherence factor plasmid pathogenicity island (locus for enterocyte effacement [LEE]) |
|EIEC ||Food, water; children in and travelers to developing countries ||Watery diarrhea, occasionally dysentery ||Invasion of colonic epithelial cells, intracellular multiplication, cell-to-cell spread ||Multiple genes contained primarily in a large virulence plasmid |
|EAEC ||?Food, water; children in and travelers to developing countries; all ages, industrialized countries ||Traveler’s diarrhea, acute diarrhea, persistent diarrhea ||Aggregative/diffuse adherence, virulence factors regulated by AggR ||Chromosomal or plasmid-associated adherence and toxin genes |
Extraintestinal Infectious Syndromes
The urinary tract is the site most frequently infected by ExPEC. UTI is an exceedingly common infection among ambulatory patients, accounting for 1% of ambulatory care visits in the United States and second only to lower respiratory tract infection among infections responsible for hospitalization. UTIs are best considered by clinical syndrome (e.g., cystitis, pyelonephritis, and catheter-associated UTI) and within the context of specific hosts (e.g., premenopausal women, compromised hosts; Chap. 130). E. coli is the single most common pathogen for all UTI syndrome/host group combinations. Each year in the United States, E. coli causes 80–90% of the estimated 6–8 million episodes of cystitis that occur in ambulatory, premenopausal women with an anatomically and functionally normal urinary tract (i.e., uncomplicated cystitis). Furthermore, 20% of women with an initial cystitis episode develop frequent recurrences.
Uncomplicated cystitis, the most common acute UTI syndrome, is characterized by dysuria, urinary frequency and urgency, and suprapubic pain. Fever and/or back pain suggests progression to pyelonephritis. Even when pyelonephritis is treated effectively, fever may take 5–7 days to resolve completely. Persistently elevated or increasing fever and neutrophil counts should prompt evaluation for intrarenal or perinephric abscess and/or obstruction. Pyelonephritis uncommonly causes renal parenchymal damage and loss of renal function, primarily in association with urinary obstruction, which can be preexisting or, rarely, occurs de novo in diabetic patients who develop renal papillary necrosis as a result of kidney infection. Pregnant women are at unusually high risk for developing pyelonephritis, which can adversely affect the outcome of pregnancy. As a result, prenatal screening for and treatment of asymptomatic bacteriuria during pregnancy are standard. Prostatic infection (prostatitis), a potential complication of UTI in men, can present in either an acute (severe) or a chronic (recurrent cystitis) manner. Acute pyelonephritis, acute prostatitis, and other systemic illnesses due to UTI can be designated collectively as urosepsis, febrile UTI, or systemic UTI. The diagnosis and treatment of UTI, as detailed in Chap. 130, should be tailored to the individual host, the nature and site of infection, and local patterns of antimicrobial susceptibility.
ABDOMINAL AND PELVIC INFECTION
The abdomen/pelvis is the second most common site of extraintestinal infection due to E. coli. A wide variety of clinical syndromes occur in this location, including acute peritonitis secondary to fecal contamination, spontaneous bacterial peritonitis, dialysis-associated peritonitis, diverticulitis, appendicitis, intraperitoneal or visceral abscesses (hepatic, pancreatic, splenic), infected pancreatic pseudocysts, and septic cholangitis and/or cholecystitis. In intraabdominal infections, E. coli can be isolated either alone or, as occurs more often, in combination with other facultative and/or anaerobic members of the intestinal microbiota (Chap. 127).
E. coli is not usually considered an important cause of pneumonia (Chap. 121). Indeed, enteric GNB account for only 1–3% of cases of community-acquired pneumonia, in part because these organisms colonize the oropharynx only transiently in a minority of healthy individuals. However, rates of oral colonization with E. coli and other GNB increase with severity of illness and antibiotic use. Consequently, GNB are a more common cause of pneumonia among residents of LTCFs and are the most common cause (60–70% of cases) of hospital-acquired pneumonia (Chap. 137), particularly among postoperative and ICU patients (e.g., ventilator-associated pneumonia).
Pulmonary infection is usually acquired by small-volume aspiration but occasionally occurs via hematogenous spread, in which case multifocal nodular infiltrates can be seen. Tissue necrosis, probably due to bacterial cytotoxins, is common. Despite significant institutional variation, E. coli is generally the third or fourth most commonly isolated type of GNB in hospital-acquired pneumonia, accounting for 5–8% of episodes in both U.S.-based and Europe-based studies. Regardless of the host, pneumonia due to ExPEC is a serious disease, with high crude and attributable mortality rates (20–60% and 10–20%, respectively).
E. coli is one of the two leading causes of neonatal meningitis, together with group B Streptococcus (See also Chap. 133). Most E. coli strains that cause neonatal meningitis possess the K1 capsular antigen and derive from a limited number of meningitis-associated clonal groups. Ventriculomegaly occurs commonly. After the first month of life, E. coli meningitis is uncommon, occurring predominantly in the setting of surgical or traumatic disruption of the meninges or hepatic cirrhosis. In patients with cirrhosis who develop meningitis, the meninges are presumably seeded as a result of poor hepatic clearance of portal vein bacteremia.
E. coli contributes frequently to infections of decubitus ulcers and occasionally to infections of lower-extremity ulcers and wounds in diabetic patients and other hosts with neurovascular compromise. Osteomyelitis secondary to contiguous spread can occur in these settings. E. coli also causes cellulitis or infections of burn sites and surgical wounds (accounting for ~10% of surgical site infections), particularly when the infection originates close to the perineum. E. coli causes hematogenously acquired osteomyelitis, especially of vertebral discs and bodies, accounting for up to 10% of cases in some series (Chap. 126). E. coli occasionally causes orthopedic device–associated infection or septic arthritis and rarely causes hematogenous myositis. Myositis or fasciitis of the thigh due to E. coli should prompt an evaluation for an abdominal source with contiguous spread.
Despite being one of the most common causes of bacteremia, E. coli rarely seeds native heart valves. When the organism does infect native valves, it usually does so in the setting of prior valvular disease. E. coli infections of aneurysms, the portal vein (pylephlebitis), and vascular grafts are quite uncommon.
E. coli can cause infection in nearly every organ and anatomic site. It occasionally causes postoperative mediastinitis or complicated sinusitis and uncommonly causes endophthalmitis, ecthyma gangrenosum, or brain abscess.
E. coli bacteremia can arise from infection at any extraintestinal site. In addition, E. coli bacteremia can arise from percutaneous intravascular devices or transrectal prostate biopsy or from the increased intestinal mucosal permeability seen in neonates and in the settings of neutropenia, chemotherapy-induced mucositis, trauma, and burns. Roughly equal proportions of E. coli bacteremia cases originate in the community and in health care settings. Isolation of E. coli from the blood is almost always clinically significant and may be accompanied by the sepsis syndrome (dysfunction of at least one organ or system) or septic shock (Chap. 297).
The urinary tract is the most common source for E. coli bacteremia, accounting for one-half to two-thirds of episodes. Bacteremia from a urinary tract source is particularly common among patients with pyelonephritis, urinary tract obstruction, or urinary instrumentation in the presence of infected urine. The abdomen is the second most common source, accounting for ~25% of episodes. Although many of these episodes result from biliary obstruction (stones, tumor) and overt bowel disruption, which typically are readily apparent, some abdominal sources (e.g., abscesses) are remarkably silent clinically and require identification via imaging studies (e.g., computed tomography). Therefore, especially given the high prevalence of asymptomatic bacteriuria among elderly and functionally compromised individuals, the physician should be cautious in attributing E. coli bacteremia to a urinary source in the absence of characteristic signs and symptoms of UTI. Soft tissue, bone, pulmonary, and intravascular catheter infections are other sources of E. coli bacteremia.
Strains of E. coli that cause extraintestinal infections usually grow both aerobically and anaerobically within 24 h on standard diagnostic media and are identified readily by the clinical microbiology laboratory according to routine biochemical criteria. More than 90% of ExPEC strains are rapid lactose fermenters and are indole-positive.
TREATMENT Extraintestinal E. coli Infections
In the past, most E. coli isolates were highly susceptible to a broad range of antimicrobial agents. Unfortunately, this situation has changed. Although geographic differences exist, in general, the prevalence of resistance is >20% for ampicillin, amoxicillin-clavulanate, cefazolin, TMP-SMX, and fluoroquinolones, even in community-acquired infections. This resistance precludes empirical use of these agents for serious infections. Travel outside of the United States, prior exposure to an antimicrobial agent, or exposure to a health care setting increases the likelihood of resistance. Fortunately, >90% of isolates that cause uncomplicated cystitis remain susceptible to nitrofurantoin and fosfomycin.
ESBL-producing strains are increasingly prevalent (8–60%), with the highest prevalences in Eastern Europe, Asia, and health care settings. A growing number of reports describe community-acquired UTIs caused by E. coli strains that produce CTX-M ESBLs. Data suggest that in some locales acquisition of CTX-M-producing, fluoroquinolone-resistant strains may result from consumption of meat products from food animals treated with third- and fourth-generation cephalosporins and fluoroquinolones. Oral treatment options for such strains are limited; however, in vitro and limited clinical data indicate that fosfomycin, pivmecillinam, and nitrofurantoin often can be used for cystitis. Carbapenems, amikacin, piperacillin-tazobactam, ceftazidime-avibactam, and ceftolozane-tazobactam are the most predictably active agents overall. Carbapenemase-producing strains are also on the rise, but prevalences in most locales are <5%. Tigecycline and the polymyxins, with or without a second agent, have been used most frequently against these XDR isolates; however, the mcr-1 polymyxin resistance gene has already been described in E. coli, and its presence calls into question the future reliability of polymyxins. Ceftazidime-avibactam is active in vitro against strains that produce serine carbapenemases such as KPC but not against those that produce metalloenzymes such as NDM-1; however, relevant clinical data are limited.
This evolving antimicrobial resistance—a source of serious concern—necessitates not only the increasing use of broad-spectrum agents for empirical therapy but also the use of appropriate narrower-spectrum agents for definitive therapy whenever possible as well as the avoidance of treatment of patients who are colonized but not infected.
INTESTINAL PATHOGENIC STRAINS
Certain strains of E. coli are capable of causing diarrheal disease. (Other important intestinal pathogens are discussed in Chaps. 128, 129, and 160–163.) At least in the industrialized world, intestinal pathogenic E. coli strains are rarely encountered in the fecal flora of healthy persons and instead appear to be essentially obligate pathogens. These strains have evolved a special ability to cause enteritis, enterocolitis, and colitis when ingested in sufficient quantities by a naive host. At least five distinct pathotypes of intestinal pathogenic E. coli exist: (1) Shiga toxin–producing E. coli (STEC), which includes the subsets enterohemorrhagic E. coli (EHEC) and the recently evolved Shiga toxin–producing enteroaggregative E. coli (ST-EAEC); (2) enterotoxigenic E. coli (ETEC); (3) enteropathogenic E. coli (EPEC); (4) enteroinvasive E. coli (EIEC); and (5) enteroaggregative E. coli (EAEC). Diffusely adherent E. coli (DAEC) and cytodetaching E. coli are additional putative pathotypes. Lastly, a variant termed adherent invasive E. coli (AIEC) has been associated with Crohn’s disease (although a causal role remains unproven) but does not cause acute diarrheal disease.
Contaminated food and water are the primary transmission vehicles for ETEC, STEC/EHEC/ST-EAEC, EIEC, and EAEC, whereas person-to-person spread (direct or indirect) is the primary transmission route for EPEC and a secondary transmission route for STEC/EHEC/ST-EAEC. Gastric acidity confers some protection against infection; therefore, persons with decreased stomach acid levels are especially susceptible. Humans are the major reservoir for such strains (except for STEC/EHEC, for which bovines are the main carriers); host range appears to be dictated by species-specific attachment factors. Although there is some overlap, each pathotype possesses a distinctive combination of virulence traits that results in a pathotype-specific pathogenic mechanism (Table 156-2). With rare exceptions, these strains are largely incapable of causing disease outside the intestinal tract. Whereas disease due to STEC/EHEC/ST-EAEC occurs primarily in high-income countries, disease due to ETEC, EPEC, and EIEC occurs primarily in low- and middle-income countries in Asia, Africa, and Latin America, and disease due to EAEC occurs globally.
SHIGA TOXIN–PRODUCING E. COLI
STEC/EHEC/ST-EAEC strains constitute an emerging group of pathogens that can cause hemorrhagic colitis and the hemolytic-uremic syndrome (HUS). In contrast to other intestinal pathotypes, STEC/EHEC/ST-EAEC causes infections more frequently in industrialized countries than in developing regions. Several large outbreaks resulting from the consumption of fresh produce (e.g., lettuce, spinach, sprouts) and of undercooked ground beef have received significant media attention. In addition, a dramatic 2011 outbreak—mainly in Germany—involved an EAEC strain that acquired a Shiga toxin–encoding phage, resulting in a novel genotype, ST-EAEC (O104:H4). This strain was transmitted to the primary cases by sprouted fenugreek seeds, with subsequent human-to-human transmission, and resulted in >4000 cases and 54 deaths.
STEC strains are the fourth most commonly reported cause of bacterial diarrhea in the United States (after Campylobacter, Salmonella, and Shigella). O157:H7 is the most prominent serotype, but many other serogroups have been described, including O6, O26, O45, O55, O91, O103, O111, O113, O121, and O145. Domesticated ruminant animals, particularly cattle and young calves, serve as the major reservoir for STEC/EHEC. Ground or mechanically tenderized beef—the most common food source of STEC/EHEC strains—is often contaminated during processing. Furthermore, manure from cattle or other animals (including in the form of fertilizer) can contaminate produce (potatoes, lettuce, spinach, sprouts, fallen fruits, nuts, strawberries), and fecal runoff from these sources can contaminate water systems. Dairy products and petting zoos are additional sources of infection.
It is estimated that <102 colony-forming units (CFU) of STEC/EHEC/ST-EAEC can cause disease. Therefore, not only can low levels of food or environmental contamination (e.g., in water swallowed while swimming) result in disease, but person-to-person transmission (e.g., at day-care centers and in institutions) is an important route for secondary spread. Laboratory-associated infections also occur. Illness due to this group of pathogens occurs both as outbreaks and as sporadic cases, with a peak incidence in the summer months.
For STEC/EHEC/ST-EAEC, production of Shiga toxin (Stx2a-g and/or Stx1a,c,d) is a critical factor for occurrence of clinical disease, as demonstrated by the 2011 ST-EAEC outbreak. The stx gene is present on chromosomally integrated prophages, and various combinations of stx types and subtypes can occur in a given strain. Shigella dysenteriae strains that produce the closely related Shiga toxin Stx can also cause hemorrhagic colitis and HUS. Stx2 (especially Stx2a,c,d) appears to be more important than Stx1 in the development of HUS. All Shiga toxins studied to date are multimers comprising one A subunit that is enzymatically active and five identical B subunits that mediate binding to globosyl ceramides, which are membrane-associated glycolipids expressed on certain host cells. As in ricin, the Stx A subunit cleaves an adenine from the host cell’s 28S rRNA, thereby irreversibly inhibiting ribosomal function (i.e., protein synthesis) and potentially leading to apoptosis.
For full pathogenicity, STEC strains require additional properties such as acid tolerance and epithelial cell adherence. Most disease-causing isolates possess the chromosomal locus for enterocyte effacement (LEE). This pathogenicity island was first described in EPEC strains and contains genes that mediate adherence to intestinal epithelial cells and a system that subverts host cells by the translocation of bacterial proteins (type III secretion system). EHEC strains make up the subgroup of STEC strains that possess stx1 and/or stx2 as well as LEE. In contrast, the 2011 ST-EAEC outbreak strain lacked LEE, yet was associated with a higher proportion of patients developing HUS (22%) than the historical average for STEC/EHEC outbreaks (2–8%). Data support the essential role of the 2011 outbreak strain’s EAEC-associated virulence factors (e.g., AAF/I fimbriae, serine proteases SigA, SepA) in adherence, increased inflammation, and disruption of the intestinal epithelial barrier, which in turn increased the systemic translocation of Stx2a.
After exposure to STEC/EHEC/ST-EAEC and a 3- to 4-day incubation period, colonization of the colon and perhaps the ileum results in symptoms. Colonic edema and an initial non-bloody secretory diarrhea may progress to the hallmark syndrome of grossly bloody diarrhea (identified by history or examination). Significant abdominal pain and fecal leukocytes are common (70% of cases), whereas fever is not; absence of fever can incorrectly lead to consideration of noninfectious conditions (e.g., intussusception and inflammatory or ischemic bowel disease). Occasionally, infections caused by C. difficile, K. oxytoca (see “Klebsiella Infections,” below), Campylobacter, and Salmonella present in a similar fashion. STEC/EHEC disease is usually self-limited, lasting 5–10 days.
An uncommon but feared complication of infection with STEC/EHEC strains is HUS, which occurs 2–14 days after diarrhea, most often in very young or elderly patients; in contrast, with ST-EAEC strains, HUS occurs more commonly among non-elderly adults, especially young women. It is estimated that in the United States >50% of all HUS cases—and 90% of HUS cases in children—are caused by STEC/EHEC. HUS is mediated by the systemic translocation of Shiga toxins. Erythrocytes may serve as carriers of Stx to endothelial cells located in the small vessels of the kidney and brain. The subsequent development of thrombotic microangiopathy (perhaps with direct toxin-mediated effects on various nonendothelial cells) commonly produces some combination of fever, thrombocytopenia, renal failure, and encephalopathy. Stx-mediated complement activation may also play a role in the development of HUS. Although with dialysis support the mortality rate of HUS is <10%, survivors often have persisting renal and neurologic dysfunction.
ETEC is a major cause of endemic diarrhea in low- and middle-income countries; it is responsible for an estimated 800 million cases annually. After weaning, children in these locales commonly experience several episodes of ETEC infection during the first 3 years of life. The incidence of disease diminishes with age, a pattern that correlates with the development of mucosal immunity to colonization factors (i.e., adhesins). In industrialized countries, infection usually follows travel to endemic areas, although occasional food-borne outbreaks occur.
ETEC is the most common agent of traveler’s diarrhea, causing 25–75% of cases. The incidence of infection may be decreased by prudent avoidance of potentially contaminated fluids and foods, particularly items that are poorly cooked, unpeeled, or unrefrigerated (Chap. 119). ETEC infection is uncommon in the United States, but outbreaks secondary to consumption of food products imported from endemic areas have occurred. A large inoculum (106–108 CFU) is needed to produce disease, which usually develops after an incubation period of 12–72 h.
After adherence of ETEC via colonization factors (e.g., CFA/I, CS), disease is mediated primarily by a heat-labile toxin (LT) and/or a heat-stable toxin (STa). Disease is less severe with strains that produce only LT. Both LT and STa cause net fluid secretion via activation of adenylate cyclase and/or guanylate cyclase C (STa) in the jejunum and ileum. The result is watery diarrhea accompanied by cramps.
LT consists of an A and a B subunit and is structurally and functionally similar to cholera toxin. Strong binding of the B subunit to the GM1 ganglioside on intestinal epithelial cells leads to the intracellular translocation of the A subunit, which functions as an ADP-ribosyltransferase. Mature STa is an 18- or 19-amino-acid secreted peptide that leads to increased intracellular concentrations of cGMP. Characteristically absent in ETEC-mediated disease are histopathologic changes within the small bowel; mucus, blood, and inflammatory cells in stool; and fever.
The disease spectrum of ETEC infection ranges from mild illness to a life-threatening cholera-like syndrome. Although symptoms are usually self-limited (typically lasting for 3–5 days), infection may result in significant morbidity and mortality (>250,000 deaths annually, mostly from profound volume depletion) when access to health care or suitable rehydration fluids is limited and when small and/or undernourished children are affected.
EPEC causes disease primarily in young children, including neonates. The first E. coli pathotype recognized as an agent of diarrheal disease, EPEC was responsible for outbreaks of infantile diarrhea (including in hospital nurseries) in industrialized countries in the 1940s and 1950s. At present, EPEC infection is uncommon in high-income countries, but among infants in low- and middle-income countries is an important cause of diarrhea (both sporadic and epidemic), often accompanied by vomiting and fever. Breast-feeding diminishes the incidence of EPEC infection. Rapid person-to-person spread may occur.
Symptoms develop after colonization of the small bowel and a brief incubation period (1 or 2 days). Initial localized adherence to enterocytes via type IV bundle-forming pili leads to a characteristic effacement of microvilli, with the formation of cuplike, actin-rich pedestals mediated by factors in the LEE. Diarrhea production is a complex and regulated process in which host cell modulation by a type III secretion system plays an important role. Strains lacking bundle-forming pili have been categorized as atypical EPEC (aEPEC); increasing data support a role for these strains as intestinal pathogens in all age groups and among HIV-infected individuals. Diarrheal stool often contains mucus but not blood. Although EPEC diarrhea is usually self-limited (lasting 5–15 days), it may persist for weeks.
EIEC, a relatively uncommon (or perhaps under-recognized) cause of diarrhea, is rarely identified in the United States, although a few food-related outbreaks have been described. In low- and middle-income countries, sporadic disease is recognized infrequently in children and travelers.
EIEC shares many genetic and clinical features, as well as a common ancestor, with Shigella. Both are intracellular pathogens whose virulence is mediated by the presence of specific factors and by the loss or inactivation of other factors (antivirulence genes), which presumably occurred during these organisms’ transition from an extracellular to an intracellular lifestyle.
Colonization and invasion of the colonic mucosa, followed by replication therein and cell-to-cell spread (in part via a type III secretion system), result in the development of inflammatory colitis. However, unlike Shigella, EIEC produces disease only with a large inoculum (108–1010 CFU) and is less virulent, typically causing only mild, self-limited (7–10 days), watery diarrhea. Onset generally follows an incubation period of 1–3 days. Occasionally, EIEC can cause a shigellosis-like (dysentery) syndrome characterized by fever, abdominal pain, tenesmus, and scant stool containing mucus, blood, and inflammatory cells.
ENTEROAGGREGATIVE AND DIFFUSELY ADHERENT E. COLI
EAEC has been described primarily in low- and middle-income countries and in young children. However, recent studies indicate that it also may be a relatively common cause of diarrhea in all age groups in industrialized countries. EAEC has been recognized increasingly as an important cause of traveler’s diarrhea. It is highly adapted to humans—the probable reservoir. A large inoculum is required for infection, which usually manifests as watery and sometimes persistent diarrhea in healthy, malnourished, and HIV-infected hosts.
In vitro, EAEC cells exhibit a diffuse or “stacked-brick” pattern of adherence to small-intestine epithelial cells. Virulence factors that probably are necessary for disease are regulated in large part by the transcriptional activator AggR. The pathogenesis of EAEC disease begins with intestinal adherence, which results from stimulation of epithelial mucus production and bacterial biofilm formation, the latter mediated by fimbriae (AAF/I-III) and possibly the mucinase Pic and dispersin. Inflammation ensues, resulting in epithelial cell exfoliation, as does intestinal secretion mediated by the enterotoxins Pet, EAST-1, ShET1, and HlyE.
Some DAEC strains are capable of causing diarrheal disease, primarily in children 2–6 years of age in some developing countries, and may cause traveler’s diarrhea. The Afa/Dr adhesins may contribute to the pathogenesis of such infections.
Acute infectious diarrhea can be classified as noninflammatory (most commonly viral) or inflammatory (usually bacterial); the latter is suggested by grossly bloody or mucoid stools or a positive test for fecal leukocytes or lactoferrin (Chap. 128). ETEC, EPEC, DAEC, and EAEC cause noninflammatory diarrhea. Identification of these agents requires specialized assays (e.g., PCR-based tests for pathotype-specific genes) that are not routinely available; however, it is rarely necessary to identify the organisms because the associated diseases are self-limited. ETEC causes the majority and EAEC a minority of cases of noninflammatory traveler’s diarrhea; here again, however, definitive diagnosis generally is not necessary for management (as discussed below). If diarrhea persists for >10 days despite treatment, Giardia or Cryptosporidium (or, in immunocompromised hosts, certain opportunistic pathogens) should be sought. The diagnosis of infection with EIEC, a rare cause of inflammatory diarrhea in the United States, also requires specialized assays.
Because of the considerable public-health importance of STEC/EHEC/ST-EAEC infections, including the threat of HUS, the CDC now recommends that all patients with community-acquired diarrhea, whether inflammatory or not, be evaluated for these pathogens by simultaneous culture (to provide an isolate for strain typing and for outbreak detection and control) and detection of Shiga toxin or its associated genes. The rationale for testing all cases of community-acquired diarrhea, regardless of clinical features, is that bloody stool and fecal white blood cells (or lactoferrin) are not reliably present with STEC/EHEC/ST-EAEC infection. In addition, the use of both tests increases diagnostic sensitivity over that with either test alone.
O157 STEC/EHEC may be identified via culture by screening for E. coli strains that do not ferment sorbitol, with subsequent serotyping and testing for Shiga toxin. Selective or screening media are not available for culture-based detection of non-O157 STEC/EHEC/ST-EAEC strains. Detection of Shiga toxins or toxin genes via DNA-based, enzyme-linked immunosorbent, and cytotoxicity assays offers the advantages of rapidity and detection of non-O157 STEC/EHEC/ST-EAEC strains. Specimens positive for toxin but culture-negative for O157 should be forwarded to the local or state public-health laboratory for specialized testing.
TREATMENT Intestinal E. coli Infections
The mainstay of treatment for all diarrheal syndromes is replacement of water and electrolytes. This measure is especially important for STEC/EHEC/ST-EAEC infection because appropriate volume expansion may decrease renal damage and improve outcome.
The use of prophylactic antibiotics to prevent traveler’s diarrhea generally should be discouraged, especially in light of high rates of antimicrobial resistance. However, in selected patients (e.g., those who cannot afford a brief illness or are predisposed to infection), the use of rifaximin, which is nonabsorbable and is well tolerated, is reasonable.
When stools are free of mucus and blood, early patient-initiated treatment of traveler’s diarrhea with a fluoroquinolone or azithromycin decreases the duration of illness, and the use of loperamide may halt symptoms within a few hours. Although dysentery caused by EIEC is self-limited, treatment hastens the resolution of symptoms, particularly in severe cases. In contrast, antimicrobial therapy for STEC/EHEC/ST-EAEC infection (the presence of which is suggested by grossly bloody diarrhea without fever) should be avoided because antibiotics may increase the incidence of HUS (possibly via increased production/release of Stx). In the treatment of HUS, plasmapheresis has no benefit and the value of inhibition of C5 (via eculizumab) is unresolved.
K. pneumoniae is the most important Klebsiella species from a medical standpoint, causing community-acquired, LTCF-acquired, and nosocomial infections. K. oxytoca is primarily a pathogen in LTCFs and hospitals. Klebsiella species are broadly prevalent in the environment and colonize the mucosal surfaces of mammals. In healthy humans, the prevalence of K. pneumoniae colonization is 5–35% in the colon and 1–5% in the oropharynx; skin is usually colonized only transiently.
Most Klebsiella infections in Western countries are caused by “classic” K. pneumoniae (cKP) and occur in hospitals and LTCFs. The most common clinical syndromes due to cKP are pneumonia, UTI, abdominal infection, intravascular device infection, surgical site infection, soft tissue infection, and secondary bacteremia. cKP strains have gained notoriety because their propensity for acquiring antimicrobial resistance determinants makes treatment challenging. Clonal group ST258, many members of which produce KPC, is undergoing international dissemination. The spread of NDM-1-producing strains from India in association with medical tourism has captured the attention of physicians and the lay press.
In addition, hypervirulent K. pneumoniae (hvKP) strains that are phenotypically and clinically distinct from cKP have emerged recently, having initially been recognized in Taiwan in 1986. Although hvKP infections have occurred globally in all ethnic groups, most cases have been reported in individuals of Asian ethnicity, mainly from the Asian Pacific Rim but also from other locales. Affected individuals often have diabetes mellitus. These demographics raise the possibility of a locale-specific distribution of the organism or an increased susceptibility of Asian hosts, especially those who are diabetic. In contrast to the usual health care–associated context for cKP infections in the West, hvKP is capable of causing serious life- and organ-threatening infections in younger, healthy individuals from the community and can spread metastatically to the eyes, central nervous system, and lungs from the primary site of infection.
hvKP infection initially was characterized and distinguished from traditional infections caused by cKP strains by its (1) presentation as community-acquired pyogenic liver abscess (Fig. 156-1, top), (2) occurrence in patients lacking a history of hepatobiliary disease, and (3) propensity for metastatic spread to distant sites (11–80% of cases). More recently, the hvKP pathotype has been recognized as the cause of a variety of serious community-acquired extrahepatic abscesses and infections without liver involvement, including pneumonia, meningitis, endophthalmitis (Fig. 156-1, middle), splenic abscess, and necrotizing fasciitis. Survivors often suffer catastrophic morbidity, such as vision loss and major neurologic sequelae.
Hypervirulent pathotype of K. pneumoniae (hvKP). Top: Abdominal CT scan of a previously healthy 24-year-old Vietnamese man shows a primary liver abscess (red arrow) with metastatic spread to the spleen (black arrow). (Courtesy of Drs. Chiu-Bin Hsaio and Diana Pomakova.) Middle: A previously healthy 33-year-old Chinese man presented with endophthalmitis. (From AS Shon et al: Virulence 4:107, 2013.) Bottom: A hypermucoviscous phenotype (which does not necessarily equate with a mucoid phenotype) has been associated with hvKP strains. A positive string test is shown. However, this test is not optimally sensitive or specific. A more sensitive and specific marker is needed.
K. pneumoniae subspecies rhinoscleromatis is the causative agent of rhinoscleroma, a granulomatous mucosal upper-respiratory infection that progresses slowly (over months or years) and causes necrosis and occasionally obstruction of the nasal passages. K. pneumoniae subspecies ozaenae has been implicated as a cause of chronic atrophic rhinitis and rarely of invasive disease in compromised hosts. K. (Calymmatobacterium) granulomatis is sexually transmitted and is the causative agent of granuloma inguinale (donovanosis) that results in chronic genital ulcers (Chap. 168). These Klebsiella pathotypes are usually isolated from patients in tropical climates and are genomically distinct from both cKP and hvKP.
Although cKP accounts for only a small proportion of cases of community-acquired pneumonia in Western countries (Chap. 121), cKP and K. oxytoca are common causes of pneumonia among LTCF residents and hospitalized patients because of increased rates of oropharyngeal colonization in such individuals. Mechanical ventilation is an important risk factor. In Asia and South Africa, community-acquired pneumonia due to hvKP is becoming increasingly common and often occurs in younger patients with no underlying disease. Klebsiella is also a common cause of pneumonia in severely malnourished children in developing countries.
As in all pneumonias due to enteric GNB, typical manifestations include production of purulent sputum and evidence of airspace disease. Presentation with earlier, less extensive infection is now more common than is the classically described lobar infiltrate, bulging fissure, and currant jelly sputum. Pulmonary infection due to hvKP that has spread metastatically (e.g., from a hepatic abscess) usually includes nodular bilateral densities, more commonly in the lower lobes. Pulmonary necrosis, pleural effusion, and empyema can occur with disease progression.
cKP accounts for only 1–2% of UTI episodes among otherwise healthy adults but for 5–17% of episodes of UTI in patients with anatomical and functional abnormalities of the urinary tract, including indwelling urinary catheter use (complicated UTI). UTI due to hvKP presents more commonly as renal or prostatic abscess due to bacteremic spread than as ascending infection from the urethra and bladder.
cKP causes a spectrum of abdominal infections similar to that caused by E. coli but is less frequently isolated from such infections than is E. coli. hvKP is a common cause of monomicrobial community-acquired pyogenic liver abscess; in the Asian Pacific Rim, it has been recovered with steadily increasing frequency over the past two decades, replacing E. coli as the most common pathogen causing this syndrome. hvKP also is increasingly described as a cause of spontaneous bacterial peritonitis and splenic abscess.
When cKP and K. oxytoca cause cellulitis or soft tissue infection, it most frequently involves devitalized tissue (e.g., decubitus and diabetic ulcers, burn wounds) and immunocompromised hosts. cKP and K. oxytoca cause some cases of surgical site infection and nosocomial sinusitis as well as occasional cases of osteomyelitis contiguous to soft tissue infection, nontropical myositis, and meningitis (during the neonatal period and after neurosurgery). By contrast, hvKP has become an important cause of community-acquired monomicrobial necrotizing fasciitis, meningitis, endophthalmitis (Fig. 156-1, middle), and abscesses within the brain, subdural space, and epidural space, particularly in the Asian Pacific Rim but also globally. Cytotoxin-producing strains of K. oxytoca have been implicated as a cause of non–C. difficile antibiotic-associated hemorrhagic colitis.
Klebsiella infection at any site can produce bacteremia. Infections of the urinary tract, respiratory tract, and abdomen (especially hepatic abscess) each account for 15–30% of episodes of Klebsiella bacteremia. Intravascular device–related infections account for another 5–15% of episodes, and surgical site and miscellaneous infections account for the rest. Klebsiella is an occasional cause of sepsis in neonates and of bacteremia in neutropenic patients. However, like enteric GNB in general, Klebsiella rarely causes endocarditis or other endovascular infections.
Klebsiellae are readily isolated and identified in the laboratory. These organisms usually ferment lactose, although the subspecies rhinoscleromatis and ozaenae are nonfermenters and are indole-negative. hvKP usually possesses a hypermucoviscous phenotype (Fig. 156-1, bottom), although the sensitivity and specificity of the string test is less than optimal. A better diagnostic test for hvKP is needed.
TREATMENT Klebsiella Infections
cKP and K. oxytoca have similar antibiotic resistance profiles. These species are intrinsically resistant to ampicillin and ticarcillin and are inconsistently susceptible to nitrofurantoin. The prevalence of resistance to amoxicillin-clavulanate, fluoroquinolones, and TMP-SMX is generally >20%. Increasing resistance is mediated primarily by plasmid-encoded ESBLs (6–70%) and carbapenemases (1–18%), with the highest prevalences in Eastern Europe and Asia and among health care–associated isolates. Furthermore, isolates of cKP that produce CTX-M ESBLs have been obtained from ambulatory patients with no recent health care contact. Oral treatment for infection due to ESBL-producing Klebsiella is more challenging than that for infection due to E. coli because of the poor activity of nitrofurantoin, the lesser activity—and perhaps lesser efficacy—of fosfomycin, and limited data on pivmecillinam. Empirical treatment of serious cKP and K. oxytoca infections with amikacin or a carbapenem may be prudent, depending on local susceptibility patterns and patient-specific risk factors.
Predictably, however, the ESBL-driven use of carbapenems has selected for strains of cKP and K. oxytoca that express carbapenemases. The limited treatment options for carbapenem-resistant Klebsiella are similar to those described for E. coli. Tigecycline, the polymyxins (e.g., colistin), and ceftazidime-avibactam are the most active agents in vitro. However, ceftazidime-avibactam is not active against metallo-carbapenemases (e.g., NDM), and resistance to polymyxins is emerging (e.g., mcr-1-mediated colistin resistance). A lethal infection due to a pan-resistant K. pneumoniae isolate has already been described in the United States. Combination therapy is often used in this setting, and consultation with relevant experts is advised.
Proteus species are part of the colonic flora of a wide variety of mammals, birds, fish, and reptiles. The ability of these GNB to generate histamine from contaminated fish has implicated them in the pathogenesis of scombroid (fish) poisoning (Chap. 451).
Proteus mirabilis causes 90% of Proteus infections, which occur in the community, LTCFs, and hospitals. Proteus vulgaris and Proteus penneri are associated primarily with infections acquired in LTCFs or hospitals. P. mirabilis colonizes healthy humans (prevalence, 50%), whereas P. vulgaris and P. penneri are isolated primarily from individuals with underlying disease. By far the most common site of Proteus infection is the urinary tract, where the principal known urovirulence factors of Proteus include adhesins, flagella, IgA-IgG protease, iron acquisition systems, and urease. Proteus less commonly causes infection at a variety of other extraintestinal sites.
P. mirabilis causes only 1–2% of UTIs in healthy women, and Proteus species collectively cause only 5% of hospital-acquired UTIs. However, Proteus is responsible for 10–15% of cases of complicated UTI, primarily those associated with catheterization; indeed, Proteus accounts for 20–45% of urine isolates from chronically catheterized patients. This high prevalence is due in part to bacterial production of urease, which hydrolyzes urea to ammonia and results in alkalization of the urine. Alkalization of urine, in turn, leads to precipitation of organic and inorganic compounds, which contributes to formation of struvite and carbonate–apatite crystals, formation of biofilms on catheters, and/or development of frank calculi. Proteus becomes associated with the stones and biofilms; thereafter, it usually cannot be eradicated without removal of the stones or catheter. Over time, staghorn calculi may form within the renal pelvis and lead to obstruction and renal failure. Although biologically plausible, clinical support is lacking for the concept that urine samples exhibiting unexplained alkalinity should be cultured, and isolation of a Proteus species (or other urea-splitting organism) should prompt consideration of an evaluation for urolithiasis.
Proteus occasionally causes pneumonia (primarily in LTCF residents or hospitalized patients), nosocomial sinusitis, intraabdominal abscesses, biliary tract infection, surgical site infection, soft tissue infection (especially decubitus and diabetic ulcers), and osteomyelitis (primarily contiguous); in rare cases, it causes nontropical myositis. In addition, Proteus uncommonly causes neonatal meningitis, with the umbilicus frequently implicated as the source; this disease is often complicated by development of a cerebral abscess. Otogenic brain abscess also occurs.
Most episodes of Proteus bacteremia originate from the urinary tract; however, intravascular devices and any of the less common sites of Proteus infection are also potential sources. Endovascular infection is rare, but when endocarditis occurs it can be persistent and destructive. Proteus species are occasional agents of sepsis in neonates and of bacteremia in neutropenic patients.
Proteus is readily isolated and identified in the laboratory. Most strains are lactose-negative, produce H2S, and demonstrate characteristic swarming motility on agar plates. P. mirabilis and P. penneri are indole-negative, whereas P. vulgaris is indole-positive. The inability to produce ornithine decarboxylase differentiates P. penneri from P. mirabilis.
TREATMENT Proteus Infections
The intrinsic resistance of P. mirabilis to tetracyclines, cefazolin, nitrofurantoin, polymyxins, and tigecycline renders treatment of XDR isolates problematic. Acquired resistance to ampicillin (prevalence range, 15–60%), fluoroquinolones (11–55%), and TMP-SMX (20–50%) is common. Ampicillin-sulbactam tends to be more active, with resistance prevalences of 6–18%. In the United States and Canada, the prevalence of ESBL production by P. mirabilis remains low (generally <5%). However, rates as high as 60% have been reported from Asia. Isolates of P. mirabilis that produce CTX-M ESBLs have been recovered from ambulatory patients with no recent health-care contact (see the section on the treatment of extraintestinal E. coli infections for treatment considerations). P. vulgaris and P. penneri exhibit more extensive drug resistance than does P. mirabilis, and induction or selection of P. vulgaris variants with stable derepression of chromosomal AmpC β-lactamase may occur. For critically ill patients, carbapenems, fourth-generation cephalosporins (e.g., cefepime), ceftazidime-avibactam, ceftolozane-tazobactam, and amikacin generally display excellent activity against Proteus species (90–100% of isolates susceptible).
ENTEROBACTER AND CRONOBACTER INFECTIONS
E. cloacae and E. aerogenes are responsible for most Enterobacter infections (65–75% and 15–25%, respectively); Cronobacter sakazakii, Cronobacter malonaticus (formerly Enterobacter sakazakii), and Enterobacter gergoviae are less commonly isolated (1% for each). Enterobacter species cause primarily health care–related infections. The organisms are widely prevalent in foods, environmental sources (including equipment at health care facilities), and a variety of animals.
These organisms colonize few healthy humans, but the percentage colonized increases significantly with LTCF residence or hospitalization. Although colonization is an important prelude to infection, direct introduction via IV lines (e.g., contaminated IV fluids or pressure monitors) also occurs. Extensive antibiotic resistance has developed in Enterobacter species and probably has contributed to the emergence of the organisms as prominent nosocomial pathogens. Individuals who have previously received antibiotic treatment, have comorbid disease, and are ICU residents are at greatest risk for infection. Enterobacter causes a spectrum of extraintestinal infections similar to that described for other GNB.
Pneumonia, UTI (particularly catheter-related), intravascular device–related infection, surgical site infection, and abdominal infection (primarily postoperative or related to devices such as biliary stents) are the most common syndromes encountered. Nosocomial sinusitis, meningitis related to neurosurgical procedures (including use of intracranial pressure monitors), osteomyelitis, and endophthalmitis after eye surgery are less frequent. Neonates (particularly those of low birth weight) are at risk for C. sakazakii infection, including neonatal bacteremia, necrotizing enterocolitis, and meningitis (often complicated by brain abscess or ventriculitis). Contaminated powdered infant formula has been implicated as a source for such neonatal infections. The WHO recommends that, to reduce the initial number of bacteria, powdered infant formula should be reconstituted with hot water (>70°C) and, to limit replication of residual bacteria, the reconstituted formula should be stored at <5°C or its storage time minimized.
Enterobacter bacteremia can result from primary infection at any anatomic site. In bacteremia of unclear origin, the contamination of IV fluids or medications, blood components or plasma derivatives, catheter-flushing fluids, pressure monitors, and dialysis equipment should be considered, particularly in an outbreak setting. Enterobacter can also cause bacteremia in neutropenic patients. Enterobacter endocarditis is rare, occurring primarily in association with illicit IV drug use or prosthetic valves.
Enterobacter is readily isolated and identified in the laboratory. Most strains are lactose-positive and indole-negative.
TREATMENT Enterobacter Infections
Significant antimicrobial resistance exists among Enterobacter strains. Ampicillin, ampicillin-sulbactam, and first- and second-generation cephalosporins have little or no activity. Extensive use of third-generation cephalosporins can induce or select for variants with stable derepression of AmpC β-lactamase, which confers resistance to these agents, to monobactams (e.g., aztreonam), and—in many cases—to β-lactam/β-lactamase inhibitor combinations. Resistance may emerge during therapy; in one study, this phenomenon was documented in 20% of clinical isolates. De novo resistance should be considered when clinical deterioration follows initial improvement, and third-generation cephalosporins should be avoided in the treatment of serious Enterobacter infections.
Cefepime is stable in the presence of AmpC β-lactamases; thus, it is a suitable option for treatment of Enterobacter infections so long as no coexistent ESBL is present. Detection of ESBLs in Enterobacter is difficult because of the presence of AmpC β-lactamase; nonetheless, their prevalence (particularly in E. cloacae) is known to be variable worldwide but is generally increasing and is now 5–50% overall. This increase is evidenced by 2014 data from the National Healthcare Safety Network, which documented resistance to third- and fourth-generation cephalosporins in 36.1% of Enterobacter isolates from central line–associated bloodstream infections in the United States. The prevalence of resistance has ranged from 15 to 40% for piperacillin-tazobactam and from 5 to 15% for colistin; it is more variable but generally higher for the fluoroquinolones. Fortunately, carbapenems, ceftazidime-avibactam, ceftolozane-tazobactam, amikacin, and tigecycline have generally retained excellent activity (90–99% of isolates susceptible). Once susceptibility data for a patient’s isolate become available, it is advisable to de-escalate the antimicrobial regimen whenever possible.