Klebsiella, Enterobacter & Others


TAXONOMY Back to top

Taxonomically, the Enterobacteriaceae remain a challenging group to define on a

phylogenetic level, as sequence divergence within the 16S rRNA gene, the most commonly

used molecular identification tool, is generally insufficient to accurately discriminate between

closely related members of this family (34, 75, 101). Sequencing of other housekeeping

genes (i.e., dnaJ, rpoB, gyrB, etc.) often yields different phylogenetic classifications

depending on their discriminatory powers. Moreover, 16S rRNA and housekeeping gene

sequencing results may not correspond with data generated by the gold standard, DNA-DNA

hybridization, or with multilocus enzyme electrophoresis. For instance, Salmonella bongori is

a distinct species by dnaJsequencing, DNA-DNA hybridization, and multilocus enzyme

electrophoresis, while tuf, atpD, and 23S rRNA sequences place S. bongori as a subspecies

within Salmonella enterica (101). Adoption of a cohesive approach to classification would

provide some stability to the taxonomy of this family. Gevers et al. (47) have suggested

using multilocus sequence analysis with a set of housekeeping genes selected on the criteria

that (i) they are useful for all strains within the taxon being studied, (ii) they present as

single copies, and (iii) recombination would not convey a selective advantage. The

continuous reclassification of organisms, particularly those with clinical significance, can

confuse clinicians, which could have serious consequences for patients.

A number of additions or taxonomic changes within the Enterobacteriaceae have occurred in

the interval since the ninth edition of this Manual; however, only two are of clinical

importance. A new species of Morganella, M. psychrotolerans, has been reported from coldsmoked

tuna and has been implicated in an outbreak of histamine (scombroid) poisoning

(38). Unlike Morganella morganii strains, which produce histamine from the histidine present

in seafood at temperatures above normal refrigeration (7 to 10°C), M.

psychrotoleransproduces toxic concentrations of histidine between 0 and 5°C, so even

properly refrigerated product could cause disease. The second change of clinical significance

is the move of Enterobacter sakazakii, a species recognized to be heterogeneous at the DNA

level, to a newly created genus, Cronobacter, with the subsequent creation of four additional

species and an unnamed genomospecies (69). Although not common,E. sakazakii is well

recognized by infectious disease clinicians as a cause of severe disease (meningitis,

necrotizing colitis, and bacteremia) with relatively high mortality rates in neonates. At this

time, species ofCronobacter cannot be routinely identified in clinical laboratories due to lack

of discriminatory biochemical tests.E. sakazakii remains an acceptable, validly published

name, but PubMed queries indicate that, based on usage,Cronobacter sakazakii is gaining

acceptance in the literature. The best options available for reporting this organism would be

to continue to use E. sakazakii or use Cronobacter sakazakii complex until such time as the

species are readily identifiable by routinely used methodologies. When laboratories adopt

taxonomic changes, the older epithet should be included in parentheses following the new

name for a minimum of 6 months (11). For infrequently encountered pathogenic organisms,

6 months may be insufficient for physicians to become familiar with the taxonomic change. It

is the laboratory’s responsibility to ensure that clinicians are aware of the significance of the

organism being reported in these cases.

Other genera, including Pantoea (Enterobacter) agglomerans and Enterobacter

cloacae remain heterogeneous at the DNA level, but because they are not phenotypically

separable, genomic groups residing within these species remain unnamed (46, 62). Hafnia

alvei is composed of two distinct DNA hybridization groups, and studies are in progress to

name DNA hybridization group 2 (78).

All the other new species of Enterobacteriaceae covered in this chapter that have been added

to or transferred between existing genera are included in Tables 1 and 2. The genera listed in

the last two columns of Table 2are not isolated from human clinical specimens or are isolated

but may not be significant, and most of these taxa will be unfamiliar to clinical

microbiologists. Genera belonging to the Enterobacteriaceae that contain one or more

species isolated exclusively from insects, plants, fish, marine animals, or birds are not

included in the tables or text of this chapter; however, information on them is available

elsewhere (72).

DESCRIPTION OF THE GENERA Back to top

Members belonging to the family Enterobacteriaceae are gram-negative, facultative

anaerobic rods or coccobacilli ranging from 0.3 to 1.0 μm wide to 0.6 to 6.0 μm

long. Serratia marcescens subsp. sakuensis is the only reportedly spore-forming organism in

this family (1). Prototrophic strains grow readily on ordinary media. Among these genera,

auxotrophic strains from clinical specimens are rare. However, cysteine-requiring urinary

isolates of Klebsiella pneumoniae, which grow as pinpoint colonies on routine media, do

occur. When encountered, these strains require supplementation of biochemical media or

commercial identification systems with 0.63 mM cysteine for accurate identification. K.

granulomatis is culturable only by cell culture techniques.

Of the organisms in Tables 1 and 2 isolated from human specimens, all Klebsiella,

Leminorella, Moellerella, Tatumella, and Enterobacter asburiae strains are nonmotile,

although any strain of any genus may be nonmotile and recent data for E. asburiae indicate

some strains may be motile (64). Some strains of Serratia plymuthicamay not grow at 37°C,

but most other members of the genera discussed in this chapter grow well between 25°C

and 37°C. Only Klebsiella and Raoultella spp. are encapsulated, but strains from all genera

may grow as mucoid or rough colonies. Six genera produce pigment. Some strains of S.

marcescens and most Serratia rubidaea and S. plymuthica strains produce a red pigment,

prodigiosin, which may appear throughout the entire colony or only as a red center or

margin. Yellow-pigment-producing organisms include environmentalEnterobacter species (E.

pulveris and E. turicensis [133, 134]) and most strains of Cronobacter (Enterobacter)

sakazakii, P. agglomerans, Leclercia adecarboxylata, and Photorhabdus asymbiotica. Yellow

pigment may be enhanced by incubation at 25°C; weak pigment producers may only be

detected by observing growth placed on a swab or filter paper. Photorhabdus

luminescens and P. asymbiotica cultures are luminescent, giving a visible glow in a darkroom

after 5 minutes. Serratia odorifera, as indicated by its name, and some Cedecea spp.

produce a pungent (potato-like) odor due to the production of alkyl-methoxypyrazines (50).

Species of Proteusand Providencia oxidatively deaminate α-amino acids, producing pyruvic

acids. L-Phenylalanine deamination yields a green color when ferric chloride is added;

however, deamination of dl-tryptophan produces the deep reddish brown pigment often seen

in media inoculated with these organisms without the addition of ferric chloride

(114). Proteus species also produce swarmer cells, i.e., elongated forms created when cells

fail to septate or divide. These cells, which are profusely covered with flagella, act in concert

to produce swarming motility on solid media (12).

Plesiomonas shigelloides organisms are also gram-negative, facultative anaerobes growing

as straight rods of sizes similar to those of other Enterobacteriaceae. However, unlike

other Enterobacteriaceae, P. shigelloidesstrains are oxidase positive, do not produce gas

from glucose (Enterobacteriaceae are variable), and are susceptible to O/129 vibriostatic

agent (2,4-diamino-6,7-diisopropylpteridine). Both P. shigelloides and enterobacteria grow at

similar salt concentrations (0 to 5%) and pH ranges (4.0 to 8.0).

EPIDEMIOLOGY, TRANSMISSION, AND CLINICAL

SIGNIFICANCE Back to top

The Enterobacteriaceae are widely distributed throughout the environment (Tables 1 and 2).

Many species of the genera in Table 1 are commonly recognized pathogens, consistently

ranking among the top 10 organisms seen in health care-associated infections (42, 61, 94).

Between 2002 and 2004, 7 of the 10 most common gram-negative organisms isolated from

respiratory tract, urinary tract, and bloodstream infections from intensive care unit patients

in the United States were K. pneumoniae (15%), E. cloacae (9%), S.

marcescens(6%), Enterobacter aerogenes (4%), Proteus mirabilis (4%), Klebsiella

oxytoca (3%), and Citrobacter freundii (2%). Between 2006 and 2007, the National

Healthcare Safety Network (formerly National Nosocomial Infections Surveillance System)

reported K. pneumoniae, Enterobacter spp., and K. oxytoca among the top 10 most

frequently isolated health care-associated infections, making up 6%, 5%, and 2% of the

isolates, respectively (61). Notably, Enterobacter was reported as the third most common

isolate from ventilator-associated pneumonia. European data on infection rates, which are

similar to those above, can be found on the following

websites: http://ecdc.europa.eu/en/Publications/AER_report.aspx andhttp://www.hpa.org.uk

/web/HPAweb&HPAwebStandard/HPAweb_C/1201767919826. Pediatric patient data collected

in 2004 from three continents (North America, Latin America, and Europe) indicated

that Klebsiellaspp., Enterobacter spp., P. mirabilis, and Serratia spp. ranked 4th, 7th, 11th,

and 12th, respectively, in the top 15 most frequently isolated organisms (42).

Both Klebsiella and Enterobacter were more prevalent (3rd to 5th versus 10th) in North and

Latin America than in Europe. In all geographic areas there was a twofold decrease in

prevalence for both Klebsiella and Enterobacter species in children older than 1 year.



Klebsiella and Raoultella

Klebsiella is carried in the nasopharynx and the bowel; however, feces are arguably the most

significant source of patient infections (98). Recent data indicate that K.

pneumoniae bloodstream infection isolation rates are 1.5 times greater during the warmest

months of the year (7). These rates most likely reflect increased fecal carriage in humans,

which in turn is a reflection of increased organisms in the environment during warm months.

This has important implications since colonized patients have a fourfold-increased risk of

infection over noncarriers. Similar isolation rate increases were not seen

with Enterobacter or Serratia. K. pneumoniae,primarily strains with capsular type K1, have

emerged as an important cause of community-acquired pyogenic liver abscess worldwide

(19, 91, 117, 146). The majority of patients with Klebsiella pyogenic liver abscess are Asian

males, 50 to 60 years of age, who present primarily with a right-lobe, solitary,

monomicrobial abscess. Studies by Brisse et al. (19) indicate that pyogenic liver abscessassociated

K1 isolates belong to a clonal complex designated CC23K1, whereas K1 strains

associated with severe pneumonia isolated from the respiratory tract and bloodstream

infections belong to a different complex, CC82K1. magA (mucoviscosity-associated gene),

which resides within the cps (capsular polysaccharide synthesis) operon of all K1 isolates, is

not an indicator for primary liver abscess. Conversely, the gene allS (an activator of the

allantoin regulon) has been detected, to date, only in K1 strains isolated from primary liver

abscesses (19, 148). allS may confer an advantage in these strains because levels of

allantoin, which can serve as a carbon source, are elevated in non-insulin-dependent

diabetics (148). Primary liver abscess K1 strains also appear to be separable by conventional

biochemical methods (see “Identification” below). Mouse lethality studies show

that Klebsiellaliver abscess isolates that possessed both the hypermucoviscosity phenotype

and the rmpA (regulator of mucoid phenotype) gene, regardless of the K type, have a 50%

lethal dose of <102 CFU versus >5 × 107 CFU for type K1 or K2 urine isolates that were not

hypermucoviscous and without rmpA or aerobactin genes (148).Klebsiella

rhinoscleromatis and Klebsiella ozaenae are clones of K. pneumoniae (not subspecies) that

have adapted to cause specific chronic infectious diseases, i.e., rhinoscleroma and atrophic

rhinitis (ozena), respectively (19). Neither organism is isolated from the environment or the

intestinal tract, and both have lost the ability to utilize substrates involved in plant product

degradation pathways; however, isolates from bloodstream, urinary tract, and other infection

sites indicate that K. ozaenae is more diverse in its ability to cause disease than K.

rhinoscleromatis. Atrophic rhinitis is restricted to the nose, but rhinoscleroma may spread to

the trachea and larynx (73). Both of these tissue-destructive diseases occur more frequently

in tropical areas of the world and are spread by person-to-person transmission, although

prolonged contact with persons producing airborne nasal secretions is required. A recent

retrospective study has updated the epidemiological and clinical features of rhinoscleroma

(35). Klebsiella granulomatis, also a clone (not species) ofK. pneumoniae, is the agent of

donovanosis or granuloma inguinale, a disease characterized by chronic genital ulcers

(19, 56). It also occurs predominantly in tropical countries and is thought to be sexually

transmitted, with humans as the only known reservoir. K. oxytoca strains carrying a

chromosomally encoded heat-labile cytotoxin have been increasingly recognized as a cause

of antibiotic-associated hemorrhagic colitis (49, 150). Antibiotic-associated hemorrhagic

colitis, associated with the use of β-lactam antibiotics, is self-limiting and resolves

spontaneously with the withdrawal of the contributing antibiotic. It differs from Clostridium

difficiledisease in that there is no pseudomembrane formation and K. oxytoca antibioticassociated

hemorrhagic colitis stools are bloody. The majority of isolations reported to date

for Klebsiella variicola are from sterile sites, mainly blood and urine (6, 121). Recent isolates

of this organism (three of five from urine), identified by rpoBsequencing reported from a

Brazilian study, provide increased evidence that this organism is a human pathogen

(6). Raoultella planticola and Raoultella terrigena share pathogenicity characteristics with K.

pneumoniae and are difficult to distinguish from it biochemically without special tests. In

European studies, 3.5% to 19% of clinical strains initially identified as Klebsiella were R.

planticola, while in U.S. and Brazilian surveys of 436 and 122 strains, respectively, only one

isolate in each was identified as R. planticola, indicating that prevalence of these species may

differ geographically (6, 113, 142).

Enterobacter, Pantoea, and Erwinia

Nosocomial Enterobacter colonization and infection are frequently associated with

contaminated medical devices and instrumentation; however, Enterobacter spp. are

commonly consumed in foods, so endogenous sources should also be considered

(73). Enterobacter ludwigii and all Enterobacter hormaechei subspecies are isolated from a

variety of human sources including blood (63, 65). A nonhuman source for E. ludwigii has

not been reported, but plants appear to be the natural habitat of E. hormaechei. Other

environmental organisms including a Pantoea dispersa-like organism, Pantoea

ananatis, and Erwinia persicina have been reported from blood or urine (33, 105, 126). P.

agglomerans is the species of Pantoea most commonly isolated from humans. Sporadic

infections are associated with penetrating trauma by objects contaminated with soil or

vegetation resulting in soft tissue infections, septic arthritis, or osteomyelitis, whereas health

care-associated infections and outbreaks often involve contaminated intravenous fluid,

parenteral nutrition, or other administered fluids.P. agglomerans infections in children, most

of whom have severe underlying conditions, are predominantly polymicrobic and therefore of

questionable significance even when specimens are from sterile sites (30).

Serratia

Serratia spp. are notorious health care-associated pathogens and colonizers. Transmission is

predominately from person to person, but medical apparatuses, intravenous fluids, and other

solutions are often implicated as well (73). Indwelling catheters, particularly for urinary tract

infections, serve as a primary reservoir for transmission via hospital personnel. In children,

the gastrointestinal tract is a common source of infections. Outbreaks transmitted by hand

are often insidious, occurring over long periods of time, and may subside and peak a number

of times before recognition and infection control efforts can contain them. Pigment

production in S. marcescens appears to be a marker that the strain is environmental in origin

and of low virulence (9). Community-acquired infections are rare except for S.

marcescens contact lens-induced acute red eye (67). Most of the other species

of Serratia have also been isolated from humans, in whom they too are usually transients or

cause opportunistic infections.

Citrobacter

Citrobacters are primarily inhabitants of the intestinal tract, and their presence in the

environment may reflect fecal excretion by humans and animals; the natural habitat of some

newer Citrobacter species is unknown. The three species most commonly involved in hospital

infections are C. freundii, C. koseri, and C. braakii; patients are usually older adults (≥65

years), more often males, and the urinary tract is the most common site of infection

(90, 123). One-third to one-half of Citrobacter infections, including septicemias, are

polymicrobic and are associated with higher mortality rates (18% to 50%) or longer hospital

stays (73, 90, 123). Meningitis is almost exclusively associated with C. koseri and involves

children <2 months of age, with the highest onset rates recorded in neonates with a mean

age of 7 days (73). Brain abscesses occur in 75% of infected infants, and neurological

defects are common sequelae in surviving infants. The most prominent risk factor is prior

colonization; during outbreaks, colonization rates of 27% have been noted versus a normal

rate of <1% (93,143). Person-to-person spread by hospital personnel and, less often, from

mother to offspring is the most likely source of infections. Sampling of inanimate or

environmental reservoirs in hospitals usually fails to

yieldCitrobacter. Other Citrobacter species including C. gillenii and C. murliniae have been

found in human clinical, animal, food, and environmental specimens (16, 18). The role of

many Citrobacter spp. in human infections is unclear because reports in the literature are

insufficient to determine either clinical significance in humans or potential reservoirs for

infection. C. rodentium causes murine colonic hyperplasia, which is self-limiting in adult mice

but causes significant morbidity and mortality in infant mice in mouse colony outbreaks (95).

Proteus, Providencia, and Morganella

Members of Proteus, Providencia, and Morganella genera are widespread in the environment,

are normal inhabitants of the gut, and are relatively common in clinical laboratories,

especially P. mirabilis. In a large six-year population-based survey of Proteeae, 85% (4,290

of 5,047) of isolates were community acquired, although providenciae were more likely to be

acquired in nursing homes (89). Females (69%) and the elderly (median age, 70) were at

highest risk of infection, and the most common specimen sources were urine (86%), soft

tissues (7%), blood (3%), miscellaneous fluids (2%), and the respiratory tract (2%). P.

mirabilis,predominantly from urine, was the most frequently isolated agent (77%). Non-P.

mirabilis species (primarilyProteus vulgaris) were isolated from wounds/soft tissue more

often than urine. Of the Providencia species, P. stuartii was isolated twice as often as other

species. P. mirabilis, Proteus penneri, Morganella, and Providencia alcalifaciens are seen in

diarrheal stools with greater frequency than in normal stools, leading to speculation that they

may cause diarrhea. Some strains of P. alcalifaciens are invasive in HEp-2 cell assays and

elicit diarrhea in the RITARD (reversible intestinal-tie adult rabbit diarrhea) model, while

other strains isolated in pure culture or in large numbers from diarrheal stools fail to invade

cell lines (4, 76). However, a number of noninvasive P. alcalifaciens strains have been shown

to be nonadherent to cell lines in vitro, which may provide insight to their inability to invade

(86). Yoh et al. (147) used a specialized medium to isolate nine strains ofProvidencia

rettgeri from 130 persons with traveler’s diarrhea; eight of these strains were invasive in

Caco-2 cells, indicating their potential for virulence in humans. Notably, vomiting was

present in five P. rettgeri cases but was not seen in patients from whom other providenciae

were isolated.

Hafnia

Few systematic investigations regarding the ecological distribution of Hafnia have been

published, although it is a common inhabitant of the gastrointestinal tract of mammals,

birds, cold-blooded animals, fish, and insects (74). Isolation from consumables, especially

meats, is not uncommon, and presumably its presence indicates prior contact of the item

with feces. Hafnia alvei has been linked to gastrointestinal disease, although putative

virulence characteristics have not been demonstrated (3, 120). However, a toxigenic strain

producing a cytopathic effect on Vero cells indistinguishable from Shiga toxin, but not

neutralized by anti-Shiga toxin antibody, has been reported (29). Although seen infrequently

in extraintestinal disease, such infections occur both in healthy and in immunocompromised

patients with monomicrobial infection rates varying from 12 to 75%; the correlation with

disease increases when the organism is isolated in pure culture and in high numbers

(74). Hafnia appears to have a predilection for the biliary tree and may produce abscesses at

the site of infection (118).

Plesiomonas shigelloides and Edwardsiella tarda

Plesiomonas is isolated from a wide range of mammals, birds, fish, water-dwelling reptiles,

and amphibians, but with the possible exception of cats, there is no evidence it plays a role

in diarrheal disease in any of these species (70). Human P. shigelloides infections are

associated with living in or travel to tropical countries and/or a history of seafood

consumption; both acute diarrhea and chronic diarrhea episodes of >2 weeks have been

reported (77, 84). Most infections are self-limited with hospitalization required only for

severe infections and/or in patients with underlying conditions. Plesiomonas typically

presents as a secretory (watery) diarrhea, and although it can also manifest as a dysenteric

(bloody) diarrhea, the secretory form is seen three times more frequently (144). Possible

diarrheal virulence determinants in some isolates include cholera-like, heat-stable, and heatlabile

toxins, while other strains have been shown to invade and multiply within human

gastrointestinal cells (71, 139). The somatic antigen of Plesiomonas may also play a role in

pathogenicity, since the gene encoding the most common type, O17, shares almost complete

identity with the form 1 (smooth) antigen gene of Shigella sonnei (25). Wound infections

associated with water contact are not encountered with plesiomonads despite their aquatic

reservoir. Plesiomonas bacteremia, which is rare and usually polymicrobic, is generally

community acquired, and major risk factors include biliary tract disease and advanced age

(>75 years) (145). Edwardsiella tarda is typically associated with water and animals that

inhabit water; it is an infrequent cause of gastroenteritis in humans, with most infections

linked to contact with fish or turtles. A low carriage rate in humans, except in tropical areas

of the world, and the ability to produce a cell-associated hemolysin and invade HEp-2 cells

suggest that E. tarda is a diarrheal agent (77). Serious wound infections, including

myonecrosis, are reported in immunocompetent individuals with aquatic exposure, but

systemic infections usually occur in patients with liver disease or iron overload conditions

(129).

Miscellaneous Enterobacteriaceae

Miscellaneous members of Enterobacteriaceae that are infrequently encountered in clinical

laboratories are primarily opportunistic pathogens in compromised patients or are present as

transients or commensals in clinical specimens (73). For some,

like Cedecea spp., Leminorella spp., Moellerella, and Tatumella, a reservoir has not been

determined because they are rarely isolated from nonhuman sources, while Ewingella,

Leclercia, andKluyvera spp. are found in a variety of foods, water, or animals (snails and

slugs) and like many enterics appear to be ubiquitous in the environment

(40, 54, 55, 59, 60, 66, 73, 137). Other genera isolated from humans have more specific

natural habitats such as Rahnella (water) or Yokenella and P. asymbiotica (insects and

infections from insect bites) (2, 41, 73, 88). There have been increased reports of clinically

significant isolations of L. adecarboxylata and Kluyvera species, all of which cannot be

covered here. Kluyvera infections occur in competent and immunosuppressed patients of all

age groups (23, 122). The organism is isolated from a broad spectrum of sources including

blood, tissue, urine, cerebrospinal fluid, and peritoneal fluid, although urinary tract and

bloodstream infections each account for approximately one-third of the infections. When

strains are determined to species level, Kluyvera ascorbata accounts for more than twice the

number ofKluyvera cryocrescens infections. Originally considered to be an opportunistic

pathogen, L. adecarboxylata is quite often isolated from polymicrobial infections in healthy

patients without underlying disease (58). This suggests that the coinfecting agent(s) might

alter the local tissue environment, allowing growth of Leclercia, or that a transfer of genetic

factors occurs, enhancing its virulence. The isolation of Leclercia in pure culture from

previously healthy persons (from a foot abscess and blood and wound caused by a

hydrofluoric acid chemical injury) in two recent reports would indicate that at least some

strains may be pathogenic (31, 58). Moellerella wisconsensis, a rarely reported organism

first isolated from stools of diarrheal patients, has been reported from an additional series of

five patients with disease ranging from self-limited acute watery diarrhea without mucus to

protracted diarrhea, lasting several weeks. No other common diarrheal bacterial pathogens

or parasites were present in the stools, and specimens taken after clinical recovery were

negative (116).

COLLECTION, TRANSPORT, AND STORAGE OF

SPECIMENS Back to top

The organisms covered in this chapter are, in general, readily isolated from clinical material,

and the principles in chapter 16 of this Manual on specimen collection, handling, and

processing are applicable. Most of these organisms will survive in culture deeps for

approximately a year. Long-term storage methods as recommended in chapter 9 work well

for these organisms. Plesiomonas cells do not survive for more than 1 to 2 months when

held at room temperature and should be maintained at -70°C. In our laboratory, strains

of Tatumella have failed to survive longer than a year even at -70°C.

ISOLATION PROCEDURES Back to top

Few of the clinically relevant strains covered in this chapter present difficulties in isolation

from sterile body sites. Isolation from nonsterile body or environmental sites may require

specialized media such as CHROMagar Orientation (Becton Dickinson, Sparks, MD) and

chromID CPS (bioMerieux, Hazelwood, MO), which perform similarly for the detection of

urinary tract pathogens covered in this chapter and can reliably replace MacConkey and

blood agars (20, 125). These media prevent swarming of Proteus and limit the spread of

mucoid colonies, which reduces overgrowth of pathogenic colonies. Additionally, colonies on

CHROMagar can be used to inoculate antimicrobial susceptibility tests directly without

subculture. CHROMagar media can also be used for specimens from other nonsterile sites;

when colony color was combined with indole, lysine, and ornithine decarboxylase tests and

serology, 98.7% (466 of 472) of the above organisms were correctly identified from nonurine

samples (109).

Both of the diarrheal pathogens covered in this chapter are easily isolated. E. tarda is a

lactose-negative, H2S- positive organism, indistinguishable from Salmonella on enteric

plating media (opaque or opaque with black centers). A positive indole reaction and failure to

agglutinate in specific Salmonella antisera separate the two

organisms. Plesiomonas produces non-lactose-, non-sucrose-fermenting colonies on enteric

plating media. It does not grow on thiosulfate-citrate-bile salts-sucrose medium, but on

cefsulodin-Irgasan-novobiocin medium, opaque colonies without a pink center (mannitol not

fermented) are suspicious for plesiomonads. Two other oxidase-positive

organisms, Pseudomonas and Aeromonas, grow on cefsulodin-Irgasan-novobiocin as well,

although Aeromonas colonies have a pink center with an opaque apron. Inositol fermentation

and a positive reaction in Moeller’s lysine, arginine, and ornithine tests will

differentiate Plesiomonas from these agents as well as other organisms.

Other Enterobacteriaceae involved in opportunistic infections and that may be isolated from a

variety of specimen types generally grow well on commonly used laboratory media (80).

Some genera are lactose or sucrose fermenters and give the appearance of normal biota on

enteric plating media, while others may produce H2S and appear Salmonellalike.

Rahnella, Ewingella, and Tatumella may require 48 hours for growth.Tatumella also

grows poorly on Mueller-Hinton agar, and a broth dilution method may be required for

susceptibility testing. K. granulomatis does not grow on conventional laboratory media but

has been grown in HEp-2 monolayers (22). Detection of Donovan bodies from tissue smears

using Giemsa or Wright stains is the method most commonly used to detect this organism.

However, these pleomorphic, bipolar staining bodies shaped like a closed safety pin are not

always present and are not reliable for diagnosis.

IDENTIFICATION Back to top

The biochemical tests most useful for separating members covered in this chapter are given

in Tables 3through 13. A table for the identification of Plesiomonas can be found in chapter

39, which describes vibrios, since these organisms most closely resemble one another

biochemically. Correct identification to species level is increasingly important in recognizing

strains that are of high risk for carrying extended-spectrum beta-lactamases (ESBLs),

cephalosporinases, or carbapenemases.



Enterobacter and Pantoea

Because of the genetic heterogeneity in several species, members of the

genera Enterobacter and Pantoeaappear to confound commercial systems more often than

other genera (73). E. ludwigii and E. hormaechei,species previously residing within the E.

cloacae complex, can be separated from E. cloacae by growth on 3-0-methyl-Dglucopyranose

and putrescine and 3-hydroxybutyrate, respectively (63, 65). By using

commercially available tests, the three subspecies of E. hormaechei can be separated with

adonitol, dulcitol, and D-sorbitol (subsp. hormaechei tests negative, positive, and negative,

respectively; subsp. steigerwaltii tests positive, negative, and positive, respectively; and

subsp. oharae tests negative, negative, and positive, respectively) (63). P. agglomerans is

very difficult to identify with either commercial systems or conventional biochemicals

(21, 103). Yellow-pigmented and lysine-, arginine,- and ornithine-negative organisms should

raise suspicion that the strain is P. agglomerans; Leclercia adecarboxylata and P.

asymbiotica also share these characteristics, but they can be separated from P.

agglomerans by positive indole and negative D-mannitol reactions, respectively. P. ananatis,

P. dispersa, and E. persicina most closely resemble P. agglomerans, but P. dispersa can be

separated from P. agglomerans by negative reactions for raffinose, salicin, and sucrose

and E. persicina can be distinguished by negative reactions for maltose and D-xylose,

respectively. P. ananatis is more difficult to differentiate, and all suspected isolates of this

organism, as with other rare Enterobacteriaceae, should be sent to a reference laboratory for

confirmation.

Serratia, Citrobacter, and Proteeae

Serratia spp. are generally easily identified except for the S. liquefaciens group; separation

of members within this group can be achieved using a combination of API 50 CH

(carbohydrate) and API ZYM (enzymatic) (bioMerieux, Hazelwood, MO) strips

(50, 73). Citrobacter spp. may be included in databases individually or by subgroups (C.

braakii-C. freundii-C. sedlakii, C. werkmanii-C. youngae, or C. koseri-C. amalonaticus);

however, subgroup identification requires further biochemical testing by standard

methodologies, and final species identification is delayed (73). A PYR (L-pyroglutamic acid,

Oxoid PYR) disk, which detects pyrrolidonyl peptidase, may be useful for separating

biochemically atypical strains of Citrobacter (positive) and Salmonella (negative) (13). Gramnegative,

oxidase- negative organisms that swarm on blood agar and appear flat with

tapered edges on MacConkey agar may be reported as Proteus. Spot indole-negative and

ampicillin-susceptible strains may be reported as P. mirabilis, while spot indole-positive,

ampicillin-resistant strains are reported as P. vulgaris (10). P. hauseri, previously a subgroup

of P. vulgaris, can be differentiated from P. vulgaris by negative salicin/esculin and trehalose

reactions (108). Proteus organisms that do not fit the above criteria must be fully identified

by commercial or conventional biochemical methods (10). Proteus species are identified with

95% to 100% accuracy by commercial systems, but Providencia identification rates vary

from 79% to 100% (108). When Providencia spp. are misidentified, they are usually

called Morganella or Proteus. Urea-positive P. stuartiimay be misidentified as P. rettgeri, or

the system may require additional tests for identification. Two-hour identification methods

misidentify M. morganii subsp. morganii about 66% of the time.

Other Enterobacteriaceae

Averyella dalhousiensis is not in commercial system databases. It may be confused with K.

ascorbata or misidentified as S. enterica in commercial systems. It shares biochemical traits

(positive for ONPG [o-nitrophenyl-β-D-galactopyranoside], malonate, potassium cyanide, and

fermentation of dulcitol and salicin) with Salmonella subspecies 2, 3, and

4. Ewingella and Tatumella are biochemically inactive, and the latter organism grows poorly

in vitro. Kluyvera can be identified only to genus level by commercial systems, but species

determination requires an ascorbate test and Irgasan susceptibility and/or gas liquid

chromatography profiles (73). P. asymbiotica is not in most commercial databases.

Molecular Identification

Information on molecular identification techniques is available elsewhere (chapter 4). For

laboratories using partial 16S rRNA (~500 bp) sequencing to identify members of

the Enterobacteriaceae, the Clinical and Laboratory Standards Institute (CLSI) provides

extremely useful information, including guidelines with suggested cutoff values for percent

identity scores and identification algorithms (27). Specifically, Table 3 of the MM18-A

guideline provides information on the usefulness of 16S rRNA for various enteric groups,

comments regarding relatedness within groups, alternative DNA targets, and indications for

identification to species level and recommendations for resolving species identification.

Suggested cutoff values help the laboratory provide clinicians with practical, recognizable

identifications of clinically significant organisms and allow the same organisms to be

identified with consistency between laboratories. Appropriate cutoff values notwithstanding,

accurate organism identification is ultimately dependent upon the availability of a reliable

database of known sequences for comparison of the sequences generated for the unknown

isolate. Both public and private databases are available, and each offers advantages (128).

An evaluation of two commercially available reference sequence libraries, MicroSeq (Applied

Biosystems, Foster City, CA) and SmartGene IDNS (SmartGene, Inc., Raleigh, NC), found

that the second had a greater diversity of sequences for comparison and provided userfriendly

software with enhancements such as the ability to add alternative gene target

databases and to store and compare previous clinical sequences (128).

TYPING SYSTEMS Back to top

The ability to trace the spread of nosocomial pathogens in outbreaks caused by

the Enterobacteriaceae has become a major responsibility for the laboratory.

Chapters 7 and 8 provide useful information on the molecular epidemiology of enteric

outbreaks. Molecular techniques, including plasmid analysis, ribotyping, pulsed-field gel

electrophoresis (PFGE), and various PCR methodologies all appear to be satisfactorily

discriminatory, with some working better for a specific genus or species than others. PCR

techniques, particularly repetitive element-PCR methods for the Enterobacteriaceae, have

proliferated at an astonishing pace and cannot be covered here more fully. To date, because

economic constraints dictate the need for a single method that is applicable for a variety of

organisms, PFGE remains the most universally accepted standardized technique for

epidemiological studies. The disadvantage of a long turnaround time (usually 4 days) has

been partially overcome by a rapid PFGE protocol that is suitable for most enteric bacteria as

well as other common clinical strains (45).

ANTIMICROBIAL SUSCEPTIBILITY Back to top

Increasing resistance in members of the Enterobacteriaceae (Table 13) has culminated with

the emergence of panresistant strains of K. pneumoniae for which there are no therapeutic

options (37, 110, 135, 136). While K. pneumoniae presents the most serious threat, K.

oxytoca, S. marcescens, Enterobacter spp., Proteus spp., andMorganella and Citrobacter spp.

have all been reported to possess one or more Ambler Class A ESBLs or carbapenemases,

Class C cephalosporinase, and/or class B metallo-β-lactamases. Frequently, plasmids

encoding these enzymes also carry resistance genes for aminoglycosides, quinolones, and

other antimicrobials (37, 102, 140). Although their clinical significance and the need to

control intra- and interhospital transmission make detection of these enzymes a priority for

the clinical laboratory, they are often unrecognized by routine susceptibility testing because

of the difficulty in detecting these enzymes and the lack of standard techniques

(110, 135, 136, 140). Both phenotypic and genotypic methods are available for detection of

ESBLs, cephalosporinases, and carbapenemases, and while the latter tests are limited to

large institutions and research laboratories, they identify specific genes and can detect lowlevel

resistance missed by phenotypic tests. General antimicrobial susceptibility and

specialized phenotypic testing procedures are discussed elsewhere in

this Manual (chapters 67 to 70) and by Patel et al. (110) and Sundin (136).

ESBL-Producing Enterobacteriaceae

In addition to screening and confirmatory methods recommended by CLSI (28), a

chromogenic agar, chromID ESBL (bioMerieux, Marcy l’Etoile, France), has been developed

for detection of ESBLs. Several studies evaluating this medium report chromID ESBL

sensitivity equal to or better than that obtained by automated or agar-based methods

(39, 48, 119). The specificity for chromID ESBL was similar to that for the comparison

methods in two studies (89% and 91%) but was only 11% in the third, causing those

authors to recommend that suspected ESBL-producing isolates on this medium be verified

with another test (39). The negative predictive value calculated in one study (>99%)

indicated that this medium can be used as an effective tool to rule out ESBL-producing

strains in clinical samples (119). chromID ESBL uses cefpodoxime as a substrate rather than

ceftazidime or cefotaxime, which was noted to increase its sensitivity and was thought to be

responsible for the greater recovery of CTX-M type ESBL-producing isolates in one of the

studies (48, 119). Strains with hyperproducing AmpC

(predominantly Enterobacter and Citrobacter spp.) and hyperproducing penicillinase (K.

oxytoca) enzymes produced false positives (strains with correct colony color but ESBL

negative on chromID ESBL) in all three studies from both the chromID ESBL agar and the

comparison methods (39, 48,119). In the two studies testing clinical samples, the recovery

rate for ESBL-producing strains was 7% and 4%, respectively (48, 119). In addition to

chromID ESBL media, Farber et al. (39) tested three Vitek 2 (bioMerieux, Durham, NC)

panels and two Phoenix (Becton Dickinson, Sparks, MD) panels with their corresponding

expert interpretation software for ESBL detection. Of the two Phoenix panels, the NMIC/ID-

70 panel slightly outperformed the best Vitek 2 panel result (AST-N041) with a sensitivity

and specificity of 84% and 75% versus 84% and 50%, respectively. In another study

comparing the Phoenix NMIC/ID-108 panel with the MicroScan (Siemens Healthcare

Systems, West Sacramento, CA) Neg BP combo panel type 30, both systems did well in

detecting ESBL-positive K. pneumoniae in clinical strains but MicroScan performed better

with various Enterobacteriaceae challenge strains (130). When ESBL-producing strains are

detected by automated systems, it is generally recommended that they be confirmed using a

manual method (39, 130,135).

Expanded-Spectrum Cephalosporin Resistance

The clinical significance of strains with chromosomal, inducible AmpC enzymes (transient

high-level production induced only in the presence of a beta-lactam) is still unclear and the

need for treatment controversial. However, the use of a beta-lactam in patients with strains

exhibiting permanent AmpC hyperproduction (due to mutations in the regulatory genes that

result in more efficient production of the enzyme) can result in treatment failure (135). The

fact that AmpC genes are now plasmid mediated in a variety of Enterobacteriaceaeadds to

the need for laboratories to recognize AmpC-positive strains, as they are an even greater

risk for transmission within the hospital. Their recognition is also complicated by the fact that

reduced susceptibility to cefoxitin, which may be used as a screen for AmpC activity, may

reflect a loss of permeability through the outer membranes as well. Incorrect reporting of a

strain as AmpC positive may result in the unnecessary use of carbapenems and the

concomitant risk of developing resistance to these drugs (136, 138). Nonetheless,

intermediate or reduced susceptibility to cefoxitin is still a useful indicator of AmpC presence,

signaling the need for further testing (136). Tan et al. (138) evaluated three methods for the

detection of AmpC-producing strains of Escherichia coli, Klebsiella, and Proteus spp. This

study found that the disk approximation test using imipenem, cefoxitin, and amoxicillinclavulanic

acid as inducing agents against ceftazidime had the least sensitivity (25%) when

compared with PCR (94%), agar dilution using cefoxitin } cloxacillin (90%), or a disk-based

test using cefpodoxime or cefoxitin } cloxacillin or boronic acid (an inhibitor of AmpC and

KPC enzymes). In the disk-based inhibitor assay cefoxitin } cloxacillin, using a zone increase

cutoff of ≥4 mm, gave the best overall sensitivity (95%). For the induction test, imipenem

performed the best of the agents tested. Overall AmpC activity was detected in 50% (127 of

255) of strains, and AmpC was found to be plasmid-borne in 94% of the 27 strains.

Carbapenem Resistance

Carbapenem resistance genes (which are defined in Table 13) may be plasmid-mediated

Class A carbapenemases, most commonly the KPC (K. pneumoniae carbapenemase) types,

Class A chromosomally encoded SME (S. marcescens enzyme), and IMI/NMC (imipenemhydrolyzing

β-lactamase/not metalloenzyme carbapenemase) types; Class B metallo-β-

lactamases (MBL), primarily the IMP (active on meropenem) type found in Asia; and Class D

OXA (oxacillin-hydrolyzing) carbapenemases (110, 136). Because carbapenem MICs, even

when elevated, appear susceptible by CLSI breakpoints, carbapenem-resistant strains are

difficult to detect. In testing for carbapenemases, the use of ertapenem as an indicator

substrate appears to be more sensitive for the detection of KPCs than imipenem or

meropenem (96, 110, 136). McGettigan et al. (96) screened 2,696 Enterobacteriaceae by

using an ertapenem disk diffusion test or Vitek 2 GN-20 card and obtained equivalent results,

detecting 85 ertapenem-intermediate or -resistant strains, of which 63 were KPC-positive K.

pneumoniae. While all of the K. pneumoniae KPC-positive isolates were confirmed by the

modified Hodge test and PCR, the four Enterobacter spp. isolates were found to be false

positives. Anderson et al. (8) found that the broth microdilution test was more sensitive than

disk diffusion, Etest, Vitek 2, and MicroScan assays for the detection of carbapenemase

resistance regardless of the carbapenem used. Ertapenem was the most sensitive substrate

for the disk test, Etest, and automated methods. While sensitive, ertapenem may be less

specific than imipenem or meropenem because it detects resistance caused by mechanisms

other than carbapenemases (8, 141). Tsakris et al. (141) detected KPC-producing K.

pneumoniae with 100% sensitivity using disks with imipenem, meropenem, or cefepime

alone and with 400 μg of boronic acid, although ertapenem gave false-positive results with

five KPC-negative, AmpC-positive isolates. An elevated MIC (≥1 or 2 μg/ml) to meropenem

or imipenem in automated assays or a zone of ≤19 mm with an imipenem disk can also be

used to screen isolates for KPCs or MBLs (8, 110). Isolates should then be tested by a

phenotypic method such as the modified Hodge assay (see chapter 70), which has been

found to be 100% sensitive and specific for carbapenemases. Unlike an inhibitor test using a

carbapenem or cefepime } boronic acid, the modified Hodge assay cannot differentiate

between KPCs and MBLs, information that is useful for infection control (110, 141). EDTA

inhibition assays can also differentiate between KPC and MBL production; Etest markets a

double-sided IP/IPI strip with imipenem on one end and imipenem-EDTA on the other end to

detect MBLs (110). CHROMagar KPC (CHROMagar, Paris, France) is another product available

for the detection of KPC-producing K. pneumoniae. When used on rectal swabs, its sensitivity

and specificity relative to PCR were 100% and 98%, respectively, compared to 93% and

96% for MacConkey agar with 10-μg carbapenem disks versus PCR (124). Overall, 41 of the

122 (34%) swabs yielded KPC-producing K. pneumoniae.

Resistance in Miscellaneous Enterobacteriaceae

Strains of P. mirabilis are resistant to nitrofurantoin but susceptible to trimethoprimsulfamethoxazole

(SXT), ampicillin, amoxicillin, piperacillin, cephalosporins, aminoglycosides,

and imipenem. Although most strains are susceptible to ciprofloxacin, resistance occurs with

unrestricted use of the drug (108). P. penneri and P. vulgarishave a resistance profile similar

to that of Morganella, although P. penneri is more resistant to penicillin than P. vulgaris. All

three organisms are susceptible to broad-spectrum cephalosporins, cefoxitin, cefepime,

aztreonam, aminoglycosides, and imipenem. They are resistant to piperacillin, amoxicillin,

ampicillin, cefoperazone, cefuroxime, and cefazolin. P. rettgeri and P. stuartii are resistant to

gentamicin and tobramycin but susceptible to amikacin. Urine isolates are susceptible to

broad- and expanded-spectrum cephalosporins, ciprofloxacin, amoxicillin-clavulanic acid,

imipenem, and SXT. Providencia heimbachae, although infrequently seen in humans, is

resistant to tetracycline, most cephalosporins, gentamicin, and amikacin. Human isolates

of E. tarda are susceptible to cephalosporins, aminoglycosides, imipenem, ciprofloxacin,

aztreonam, and antibiotic-β-lactamase inhibitor combination agents (26). Isolates from fish

and fish ponds may be more resistant because of the antibiotics used prophylactically in fish

farming. Most strains of E. tarda produce β-lactamases, even though they are susceptible to

β-lactams. P. shigelloides is resistant to ampicillin, carbenicillin, piperacillin, and ticarcillin

and is variably resistant to most aminoglycosides and tetracycline (71). Cephalosporins,

quinolones, carbapenems, and SXT show good activity against P. shigelloides.

Susceptibility results of uncommonly seen species of Klebsiella,

Enterobacter, and Serratia are similar to those of conventional species within these genera

(44). Susceptibilities for other Enterobacteriaceae vary from isolate to isolate, so that no

empirical guidelines are available for therapy prior to susceptibility testing of the suspected

strain.

EVALUATION, INTREPRETATION, AND REPORTING OF

RESULTS Back to top

When commercial systems identify species included in this chapter with a high level of

accuracy (>90% probability), the identification is probably reliable. However, for organisms

isolated from sterile sites that are identified with a probability of <90%, the isolate should be

confirmed by conventional or molecular methods or sent to a reference laboratory using

these techniques. In the interim, the isolate may be reported to the physician with a

presumptive identification. Rare species that are identified with low probabilities should

always be sent to a reference laboratory accompanied by a brief history.

An increasing number of patients are infected or colonized

with Enterobacteriaceae possessing ESBLs and carbapenemases, and they serve as

reservoirs for transmission of these enzymes within and between health care institutions.

The ability to recognize these strains is critical not only for patient care but also to prevent

the emergence of panresistant strains. Reliable, cost-effective methodologies for the

detection and reporting of these enzymes that are manageable for all institutions regardless

of size and capability are urgently needed. At the very least, all strains of K.

pneumoniae should be tested for ESBLs, since automated testing methods seem to be

reliable with this agent (39). Any strain of Enterobacteriaceae that has been shown by

susceptibility testing to have an ESBL or AmpC cephalosporinase should be reported as

resistant to all penicillins, expanded-spectrum cephalosporins, and aztreonam (136, 140).

Data provided by a CDC database of PFGE patterns used to monitor KPC-producing strains

isolated in the United States and other areas worldwide indicate that infection control

(http://www.cdc.gov/ncidod/dhqp/gl_isolation.html) interventions will be necessary to halt
the spread of these organisms (110).

No comments:

Post a Comment