Legionella

TAXONOMY Back to top

The Legionellaceae are composed of a single genus, Legionella, and 52 validly named species

(http://www.bacterio.cict.fr/l/legionella.html) (Table 1). Legionella pneumophila, L.

micdadei, L. longbeachae, andL. dumoffii are the most important from a clinical standpoint,

with L. pneumophila causing more than 90% of cases of Legionnaires’ disease (LD).

The Legionellaceae are most closely related to the Coxiellaceae, and these two families

comprise the proposed order “Legionellales” within the class Gammaproteobacteria and

phylumProteobacteria phy. nov. Coxiella burnetii, the agent of Q fever, shares many

characteristics with L. pneumophila, including intracellular parasitism and close homologies

of several virulence genes (122). Some investigators proposed the use

of Tatlockia and Fluoribacter as additional genera within the Legionellaceae (47), but a

subsequent study of 16S rRNA demonstrated that the Legionellaceae are monophyletic (44);

use of these genus names was never widely accepted and is of historical interest only. A

number of Legionella-like bacteria have been described to grow only within free-living

amoebae, and have been designated LLAP, for Legionella-like amoebal pathogen. One of the

LLAPs, Legionella lytica has been shown to cause human disease. Four have been assigned

novel Legionella spp.; three of these four have been grown axenically at low temperature

(1).

Four different L. pneumophila serogroup 1 strains have been completely sequenced: the

Philadelphia 1, Lens, Paris, and Corby strains. Analysis of these sequences shows that L.

pneumophila contains a number of eukaryotic-like genes, some of which have been shown to

allow the bacteria to create bacteria-friendly intracellular environments by subverting the

normal cellular machinery (11, 12). Analysis of the L. pneumophilagenome also shows that it

is a genetically diverse species and that some virulent strains are disseminated worldwide

(11). Genes encoding for the lipopolysaccharide core region and an O side chain predominate

in clinical isolates of L. pneumophila serogroup 1 and may be the reason why this serogroup

predominates as a cause of LD (11).

DESCRIPTION OF THE AGENT Back to top

The Legionellaceae are a diverse group of mesophilic, motile, asaccharolytic, obligately

aerobic, nutritionally fastidious gram-negative rods, sharing common growth dependence

for L-cysteine, growth enhancement by iron, and cellular branched-chain fatty acids and

ubiquinones that are unusual for gram-negative bacteria (29).L. pneumophila is the most

extensively studied of the Legionella spp., with relatively little known about most of the

other Legionella spp. Almost all of the Legionella spp. have been isolated from aqueous

environmental sources, and about a third of the 52 validly named species have been isolated

from both humans and the environment. It is assumed that the natural reservoir of all

the Legionellaceae is our aqueous environment and that humans are an accidental host of

the bacterium. Environmental L. pneumophila is a facultative intracellular parasite of several

different free-living amoebae, such as Acanthamoeba and Naegleria, existing in microbial

consortia in biofilms and free-flowing water.

L. pneumophila grows at a temperature range from 20 to 42°C, with optimal growth

occurring at temperatures of 35 to 37°C. Growth on solid media is enhanced by increased

humidity. Incubation in 2 to 5% CO2 can enhance the growth of some Legionella spp.

Bacterial phenotype, including immunogenicity, cell size, and virulence, can be altered by

growth at different temperatures (29).

Amino acids, rather than carbohydrates, are used as energy sources by

the Legionellaceae growing in vitro; this is true for intracellular bacteria as well, despite the

presence of putative carbohydrate utilization genes in the L. pneumophila genome (29).

Primary isolation of all known Legionella spp. requires medium supplementation with Lcysteine,

as does successful propagation of all but a few species. Iron supplementation of

growth media is required for optimal growth, although many Legionella spp. can grow, albeit

poorly, in the absence of the mineral. Growth of L. pneumophila is enhanced by the addition

of α-ketoglutarate (0.1%) to media, via an unknown, nonnutritive, mechanism.

Growth of L. pneumophila in artificial media can be inhibited by a number of factors. These

include the presence of high (100 mM, or 0.6%) NaCl concentrations, toxic peroxides,

products of other bacteria and fungi, and some lipids (29). In addition, optimal growth

occurs over a very narrow pH range from 6.7 to 6.9. Solid growth media contain activated

charcoal to inactivate toxic lipids and peroxides and an organic buffer (MOPS

[morpholinepropanesulfonic acid] or ACES [N-(2-acetamido)-2-aminoethanesulfonic acid]) to

reduce sodium content and provide the required pH. Preparation of growth media

for Legionella spp. can be complex and is usually best left to competent commercial sources

or to specialized laboratories.

EPIDEMIOLOGY AND TRANSMISSION Back to top

LD was first recognized as a distinct entity when epidemic pneumonia with a 15% fatality

rate developed during and after a convention of the Pennsylvania American Legion in

Philadelphia in July 1976 (30, 41). Joseph McDade and colleagues at the CDC determined

that a novel gram-negative bacterium was the cause of the outbreak (83). Neither the

disease nor the bacterium was found to be novel, with the first known epidemic of LD having

occurred in 1957 (92) and isolation of the bacterium having occurred multiple times from the

1940s on.

Environmental studies found that the bacterium was widespread in natural bodies of water

and occasionally in high concentration in warm waters found in plumbing systems, water

heaters, warm water spas, and cooling towers. Many different Legionella spp. exist in nature

within free-living amoebae, and as a result, these otherwise fastidious bacteria can multiply

within the amoebae and be protected from biocides (102).Legionella-infected amoebae are

often found in complex consortia of microorganisms within biofilms. The bacteria are present

in very low concentrations in freely flowing cold water and biocide-treated waters but can

multiply in warm and, especially, stagnant water. Devices that aerosolize these contaminated

waters serve to disseminate the bacteria.

Legionella pneumophila serogroup 1 causes 95 to 98% of community-acquired LD. The

Pontiac/MAb 3-1 monoclonal antibody subgroup of L. pneumophila serogroup 1 constitutes

about 80 to 90% of clinical isolates of this serogroup. Just a few clonal types of L.

pneumophila serogroup 1, the Pontiac subgroup, are responsible for about 50% of sporadic

community-acquired infection (11, 49, 55). Several L. pneumophilaserogroup 1 strains that

predominate as clinical isolates are uncommon in the environment, whereas most L.

pneumophila strains that are commonly found in the environment are unusual causes of LD

(20, 55, 74). Infection caused by L. pneumophila serogroup 1 Pontiac subtype is less

common in nosocomial LD, especially that involving immunocompromised patients. Up to

60% of nosocomial LD may be caused by other L. pneumophila serogroup 1 subtypes,

other L. pneumophila serogroups, and other Legionella species (59, 60).

Infection is acquired by aerosol inhalation of contaminated water, although microaspiration

may also be a mechanism of acquiring the disease (30). The majority of community

epidemics of LD are from Legionella-contaminated cooling towers or other aerosol-generating

devices. Contaminated potable water systems, such as water heaters and warm water in

pipes can also be a major source of disease, although these sources are not usually the

cause of explosive outbreaks of the disease.

Despite the ubiquity of Legionella spp. in our environment, LD is an unusual cause of

pneumonia. About 0.5 to 5% of adults requiring hospitalization for pneumonia have LD.

Passive reporting indicates that the disease incidence is from 4 to 20 cases per million people

per year, and a prospective study estimated that the disease incidence is about 80 cases per

million people per year, or between 8,000 to 18,000 LD cases annually in the United States

(82). This rate may be an underestimate, as a recent German study found that the annual

rate was 180 to 360 cases/million population (117). Underreporting of LD is common in the

United States, with only about 2,800 cases reported to the CDC in 2008. Sporadic

community-acquired LD is much more common than epidemic-associated disease, in a ratio

of about 4:1.

The incubation period of LD is estimated to be between 2 to 14 days, with a median value of

about 4 days. A study of a large outbreak extended the incubation period to as long as 19

days, with a median value of 7 days (16).

L. pneumophila causes disease by infecting human mononuclear cells, primarily alveolar

macrophages. After the bacterium is inhaled into the lungs, it invades lung macrophages and

multiplies in them. Detailed descriptions of pathogenesis can be found elsewhere

(2, 30, 46, 67).

CLINICAL SIGNIFICANCE Back to top

LD is a type of bacterial pneumonia, caused by L. pneumophila and other Legionella spp. The

pneumonia ranges in severity from mild to fatal, with an average fatality rate of 12% (5).

Major risk factors for the disease include immunosuppression of the cellular immune system,

cigarette smoking, overnight travel outside the home, use of well water, chronic heart or

lung disease, and chronic renal failure. Solid-organ transplant patients are at particularly

high risk, as are patients receiving anti-tumor necrosis factor therapy for a variety of

autoimmune diseases (30).

LD cannot be readily distinguished from other forms of community-acquired pneumonia by

clinical, roentgenographic, or nonspecific laboratory studies (30). Several attempts at

developing a clinical scoring system to distinguish LD from other pneumonias have failed.

The severity of pneumonia at presentation, underlying diseases, and promptness of specific

antibiotic therapy are important prognostic factors. Promptly treated LD can be cured in 95

to 99% of cases in otherwise healthy persons. Less than half of patients may respond if

there is a delay in therapy, immunosuppression, or respiratory failure (30). Untreated

disease causes death in about 15% of previously healthy patients and up to 75% of severely

immunocompromised ones (30).

Prospective, randomized controlled studies of adequate size have not been performed to

determine the optimal therapy for LD, so great reliance is placed on experimental tissue

culture and animal model studies, as well as results of nonrandomized studies (23, 30).

Erythromycin, clarithromycin, azithromycin, a tetracycline, telithromycin, levofloxacin,

ciprofloxacin, and moxifloxacin all appear to have roughly equivalent efficacies for

nonimmunocompromised outpatients with mild LD (30). The quinolone antimicrobials,

especially levofloxacin, and azithromycin are the drugs of choice for severe disease and for

immunocompromised patients (30, 87). Antimicrobial therapy with more than one agent is

sometimes used but is of questionable benefit and, in the case of rifampin, may be harmful

(52).

Pontiac fever is an acute influenza-like illness that has been associated with exposure

to Legionella sp.-containing environmental aerosols (40, 51). The etiology and pathogenesis

of this disease are unknown, but it appears as if the disease is caused by inhalation of

bacterial toxins, such as endotoxin, or perhaps an acute allergic reaction to a bacterium.

Since multiple microorganisms, and endotoxin, have been found in aerosols causing Pontiac

fever, it is unclear if Legionella spp. play any role at all in disease causation. Pontiac fever is

self-limited, with no reported deaths, little to no need for hospitalization, and no need for

antibiotic therapy.

COLLECTION, TRANSPORT, and STORAGE OF

SPECIMENS Back to top

Expectorated sputum and other lower respiratory specimens are the most common sources

of Legionella spp. Other, less common sources include pleural fluid and blood. Rare sources

have included pericardial fluid, kidney, liver, spleen, myocardium, respiratory sinuses, skin

and soft tissues, infected wounds, peritoneal fluid, prosthetic heart valves, bone marrow, and

intestine. Culture of available sputum, bronchoscopy specimens, lung biopsy specimens, and

pleural fluid should be routine for laboratory diagnosis of LD. Lung biopsy specimens have

the highest yield but may be negative. Culture of expectorated sputum or other lower

respiratory tract secretions, second in yield to lung biopsy, should always be performed for

optimal detection of legionella infection. Pleural fluid has low yield but should be cultured if it

is available. Routine culture of other specimens for Legionella spp. is not indicated unless

there is a high clinical suspicion of the disease affecting these sites.

Sputum microscopic scoring criteria cannot be used to determine which sputum specimens

should be cultured for legionella bacteria because of limited purulence and scanty secretions

in patients with LD. Up to 80% of specimens culture-positive for Legionella spp. may be

rejected using the criteria of the presence of sputum purulence for processing specimens

(39, 66).

Urine for antigen detection should be collected in a sterile container (27). Boric acid

preserves the antigen, but use of commercial urine transport systems containing boric acid

have not been studied for antigen preservation and freedom from interactions. The urine can

be transported to the laboratory at room temperature if no more than a several-hour delay is

anticipated. Longer transport times require specimen refrigeration; urine specimens should

not be frozen, as this may reduce test sensitivity and specificity.

Blood for serum antibody testing is collected in standard tubes and transported at room

temperature (26). Test performance is not adversely affected by storage of the clotted,

unseparated blood at room temperature for several days. Long-term storage is at −20°C in

aliquots to allow parallel testing without freeze-thawing, which can lower antibody levels.

Legionella spp. are hardy and generally survive for up to a week in clinical specimens.

Sputum and other respiratory tract specimens, including lung biopsy specimens, should be

collected in sterile containers and transported to the laboratory promptly at room

temperature. Transportation and storage should be at 2 to 5°C if more than a several-hour

delay is anticipated before the specimen can be plated. Very long term storage is best at

−70°C, although this can reduce bacterial concentration to below the level of detection when

the starting concentration was low or the specimen was primarily aqueous. Repeated freezethawing

is harmful to the bacteria. Some tissues, especially spleen, contain growthpreventing

substances and must be plated promptly, since even overnight storage at 5°C

dramatically reduces culture yield; note that this is not true of lung specimens.

DIRECT EXAMINATION Back to top

Microscopy

The morphology of L. pneumophila found in lung and sputum specimens is a small

coccobacillus to short rod, 3 to 5 μm in length (Fig. 1). This is much different from that

observed for the bacterium taken from a culture plate, which is usually a long, filamentous

bacillus, 10 to 25 μm in length. L. pneumophila is very difficult to detect by Gram staining of

sputum or lung biopsy specimens. Use of 0.1% basic fuchsin, rather than safranin, greatly

enhances the staining of the bacterium from culture plates, but even with use of this stain, it

is very difficult to visualize the bacterium in sputum and tissues. Less than 0.1% of L.

pneumophila present in lung tissue or sputum can be visualized by Gram stain using basic

fuchsin. The small size of intracellular L. pneumophila, the form present in human tissues,

makes visualization difficult with Gram stain, as does stain uptake by the surrounding

proteinaceous material found in sputum and tissue specimens.

detection of L. pneumophila in embedded tissues as well as sputum. These silver stains are

useful for detection of the bacterium in embedded tissues but have no present role in the

staining of the bacterium in sputum or other nonembedded specimens. Silver stains are not

highly sensitive, can produce artifacts, and require expert use and interpretation for optimal

sensitivity and specificity.

Some Legionella spp., in particular L. micdadei, may stain with acid-fast stains, both in fresh

specimens and from Formalin-fixed tissues (8, 63). The small coccobacillary morphology

of Legionella spp. should be a clue that the acid-fast organism is not a mycobacterium.

Immunofluorescent microscopy is the most sensitive and specific microscopic method for the

detection of L. pneumophila in tissues and sputum (25, 34). Optimal sensitivity and

specificity require exacting staining methods and great expertise by the microscopist. Even

when well performed, the test sensitivity has been low compared to other diagnostic

methods (103). For these reasons, this test is now rarely used for direct examination.

Detailed discussions of this test can be found elsewhere (25).

Antigen Detection

LD due to L. pneumophila serogroup 1 can often be diagnosed by detection of bacterial

antigenuria (27). Several immunoassays are commercially available for this purpose, the

most convenient of which is a rapid single test immunochromatographic card assay.

Immunochromatographic card assays are made by at least three companies, and two are

FDA cleared (Binax, Scarborough, ME, and SA Scientific, San Antonio, TX). Of these, only the

Binax NOW assay has been extensively evaluated. Several other assays utilize a microtubebased

enzyme immunoassay. Only one non-FDA-cleared kit (Biotest, Dreieich, Germany) is

designed to detect non-serogroup 1 L. pneumophila, but the kit appears to be no more

sensitive than the other available kits (27). The strength of all these assays is their detection

of L. pneumophila serogroup 1 infections and, in particular, its Pontiac/MAB2/MAB3-1

monoclonal subtype (59–61). The immunochromatographic card assay may be somewhat

less sensitive than the microtube-based immunoassays, but its convenience, ease of use,

and robustness make up for a slight decrease in sensitivity.

Clinical test performance for all assays is dependent on the pretest probability of L.

pneumophila serogroup 1 and on the probability of Pontiac monoclonal subtype L.

pneumophila serogroup 1 infection (59, 61). The assays detect about 60 to 70% of L.

pneumophila serogroup 1 Pontiac monoclonal subtype epidemic infections and up to 90% of

sporadic pneumonia caused by this subtype. The differences in test sensitivity for the same

bacterial subtype are probably due to differences in disease severity, the other major factor

determining test sensitivity. Patients with severe L. pneumophila serogroup 1 LD are the

most likely to have positive urine antigen tests, for example, those requiring intensive care

nursing and ventilator assistance; test sensitivity in this population is probably in the range

of 90 to 95% of those infected with the Pontiac monoclonal subtype. On the other hand,

urine testing may detect only 50% of outpatients with mild epidemic disease caused by the

same monoclonal subtype, perhaps 40% of hospitalized patients with other L.

pneumophila serogroup 1 subtypes, and fewer than 5 to 40% of those with infections caused

by other serogroups and species (60, 64). The test may be negative during the first day of

illness, but those with severe disease are likely to be positive upon presentation to the

hospital. Repeat testing 2 to 3 days after the onset of illness may detect a small number of

patients who had negative tests initially.

Test sensitivity can be enhanced by concentrating the specimen using ultrafiltration devices

such as Amicon concentrators (Millipore, Billerica, MA). In some studies, this has increased

test sensitivity by about 30%, without affecting specificity (21, 54). Prolonging incubation

time for the Binax NOW assay to 60 minutes also increases sensitivity without decreasing

specificity (19, 60). Test sensitivity decreases when specimens are frozen for weeks to

months before testing.

The urine antigen assays are very specific, in the range of 99 to 99.9%. False-positive tests

can be due to urine rheumatoid-like factors, freeze-thawing of urine, and excessive urinary

sediment. All together, these causes of false-positive tests account for no more than a few

percent of all positive tests. Regardless, all positive tests should be confirmed after boiling

urine clarified by centrifugation.

Molecular Diagnosis of LD

Nucleic acid-based detection of Legionella spp. in sputum, urine, and blood samples has been

successfully used in reference and research laboratories, with detection of L.

pneumophila being the most extensively studied. The best results show that molecular

diagnosis is a more sensitive method of diagnosis than culture, although some studies

showed rough equivalence (86). Test sensitivities have been estimated to be 80 to 100%, 30

to 50%, and 50 to 90% for lower respiratory secretions, serum, and urine analytes,

respectively; test specificities are estimated to be >90% (86). Both conventional and realtime

assays have been utilized (62, 86). Most laboratories use the macrophage infectivity

protein (mip) gene target to detect L. pneumophila.Legionella spp. are most commonly

detected using an rRNA target, usually 16S, although 23S is claimed to have advantages

(89). Sputum digestion may be important to increase yield (4). Multiplex assays are most

commonly used, with no disadvantage over uniplex methods, although one study showed the

superiority of a nucleic acid sequence-based amplification uniplex assay over a multiplex

format (79, 107). Until recently, only home brew assays were available.

Three commercial assays exist, with only one cleared for marketing by the U.S. FDA (BD

ProbeTec ETLegionella; BD Diagnostics, Sparks, MD); there are no published evaluations of

this product. The other two assays, Chlamylege (Argene, North Massapequa, NY) and

Pneumoplex (Prodesse, Waukesha, WI), performed quite well in single published evaluations

(48, 70).

The added benefit of nucleic acid amplification-based detection of Legionella spp. over that

obtained by urine antigen testing appears to be slight, with 11% greater yield than urine

testing alone. This is likely because of the predominance of L. pneumophila serogroup 1,

Pontiac subgroup, in community-acquired disease (18). Performance of nucleic acid testing

for the detection of nosocomial LD and LD in immunocompromised patients would probably

be significantly higher than urine antigen testing.

Because Legionella spp. are commonly found in water, contamination of almost any

molecular reagent withLegionella spp. nucleic acid is a concern. False-positive PCR tests

for Legionella spp. have been attributed to contaminated commercially produced “pure”

water and nucleic acid extraction columns (35, 104, 114). Since sequencing of false-positive

products yielded Legionella spp. sequences, contamination cannot be excluded by the ability

to sequence the product and assign it to a particular Legionella sp. (35). In the case of

extraction column contamination, only a few columns of the same lot may be contaminated

and not the entire lot. This means that multiple negative controls are required for optimal

specificity, including extraction controls as well as no-template controls.

ISOLATION PROCEDURES Back to top

Specimen Plating

Optimal yield of Legionella spp. from clinical specimens usually requires that specimens be

diluted to reduce inhibition by tissue and serum factors as well as antibiotics, that the

specimen be pretreated to reduce contaminating microbiota, and that a variety of selective

and nonselective media be used (Table 2). Culture ofLegionella spp. from normally sterile

fluids and tissues, such as pleural fluid, aseptically obtained lung tissue, or blood, is often

successful without the use of multiple selective media and specimen decontamination.

Dilution (1:10) in tryptic soy broth increases the culture yield of most specimen types,

including sputum and other liquid respiratory tract specimens, lung tissue, lymph nodes and

spleen, and probably other organs such as liver and kidney. Sputum and other respiratory

specimens should first be examined in a Petri dish for purulent-appearing material, and this

material should be selected for culture. Tissues (about 1 g) are ground in a tissue grinder

with a small amount (1 ml) of broth, which adequately dilutes most tissues except for

spleen; this tissue requires an additional 1:10 dilution for the best recovery of bacteria.

Liquid specimens are roughly diluted by adding about 0.1 ml of vortex-mixed liquid

specimens to 0.9 ml of the dilution broth. Pleural fluid, joint fluid, and blood subcultured

from blood culture bottles do not require dilution before plating; in fact, pleural fluid yield

may be enhanced by concentration by centrifugation.

Decontamination is required to reduce contaminating microbiota in most sputum and other

respiratory tract secretions. This is done by diluting (1:10) the specimen in a low-pH KCl-HCl

buffer (pH 2.2) and incubating it at room temperature (4.0 minutes) before plating the

suspension onto culture media. Timing is critical here, with resultant low yield if the timing is

off by as little as a minute. The culture medium is sufficiently buffered so that the acidified

specimen is neutralized upon being plated. An alternative to specimen acidification is heating

at 50°C for 30 minutes. Most aseptically collected tissue specimens do not require

decontamination, although occasionally, lung tissues contain multiple contaminating bacteria

and fungi. In this case, heat or acid treatment of tissue ground in sterile distilled water may

help; sometimes dilutions of the ground tissues are also required for optimal yield, with or

without pretreatment.

Inoculation of Plates

Approximately 0.1 ml is inoculated to each plate, with the bulk of the inoculum applied to the

first quadrant. Comparative studies are lacking to show whether it is better to streak plates

for isolation or to uniformly distribute the inoculum over the entire plate. The plates must be

thoroughly dry before being inoculated to aid in absorption of the relatively large volume

inoculum and to retard spreading of contaminants throughout the plate.

Culture Media

Buffered charcoal-yeast extract (BCYE) medium supplemented with 0.1%α-ketoglutaric acid

(BCYEα), is used for isolation and growth of Legionella spp. Use of BCYE without α-

ketoglutarate supplementation cannot be recommended for clinical use, as this amino acid

greatly enhances growth of the bacterium (32).

BCYEα can be made selective by the addition of antimicrobial agents (Table 2). A variety of

different antifungal agents are used in the media. Cycloheximide is a poor choice for media

used for clinical specimens, as it fails to inhibit Candida albicans. Both anisomycin and

natamycin inhibit more yeasts than does cycloheximide. An array of media (Table 2) exists

because no one selective medium is best for all purposes. Optimal yield ofLegionella spp.

from clinical specimens requires the use of three different media, one nonselective plate

(BCYEα) and two selective media (BMPA [BCYE containing cefamandole, polymyxin B, and

eithes anisomycin or natamycin] and PAV [BCYE containing polymyxin B, vancomycin, and

either anisomycin or natamycin]). BMPA is an excellent selective medium for the vast

majority of L. pneumophila strains, but the cefamandole present in the medium inhibits the

growth of some other Legionella spp. and, rarely, L. pneumophila strains. Use of the lessselective

medium, PAV, is required for optimal growth of some Legionella spp. other than L.

pneumophila. No selective medium inhibits multiresistant gram-negative bacteria, reducing

culture yield in nosocomial disease.

Selective and nonselective media are optimized for the isolation of L. pneumophila, and their

performance for the isolation of other Legionella spp. is not accurately known. One study

showed that L. micdadei in a guinea pig spleen had enhanced recovery on BCYEα medium

prepared with 1% bovine serum albumin; the growth was enhanced because of less growth

inhibition by spleen tissue (84). Whether addition of bovine serum albumin to BCYEα medium

enhances L. micdadei recovery from human lung or sputum is unknown and probably

unlikely. A BCYEα-based selective medium containing natamycin, aztreonam, and

vancomycin has been reported to be useful for the isolation of L. longbeachae from soil

(105).

Medium shelf life is around 1 year for nonselective plates and slants. This long shelf life

requires thick plates (25-ml pour), complete drying of plates before storage at 2 to 4°C in

sealed plastic bags, and protection from light. Selective media lose selectivity after about 3

months of storage time, but depending on the incorporated antibiotic, the media may last

considerably longer.

Quality control (QC) testing of media is required before they are put into use. Current CLSI

standards are inadequate for proper QC testing of these media. About 1% of commercial

media fail laboratory QC testing (personal observations). The CLSI QC testing protocol

utilizes a heavy inoculum of medium-adapted Legionellasp. strains and a growth/no-growth

test. Minor variations in the manufacture of media, such as the addition of excess salt,

overlong autoclaving, and degradation of buffers can all seriously affect the ability of the

medium to support wild strain growth but not necessarily that of medium-adapted strains.

The optimal method for medium QC testing is the inoculation of the test media with several

hundred nonartificial medium-passed L. pneumophila bacteria (obtained from an infected

guinea pig lung) and quantification of the bacterial colonies after 3 to 4 days of incubation

(24). In the absence of the availability of lung-passaged L. pneumophilabacteria, lowpassage

clinical strains should be used, taking care to plate only several hundred bacteria

per plate. QC testing of selective media for the ability to suppress non-Legionella bacteria

can be done by inoculation of the plate with relatively antibiotic-susceptible Escherichia

coli and Staphylococcus aureus, such as ATCC 25922 and 25923; the growth should be

markedly suppressed.

Medium Incubation

Inoculated media are incubated at 35 to 37°C in humidified air. Regardless of the

humidification method, care must be taken to keep the incubators or jars very clean and to

regularly sterilize the containers or incubators. A small amount of CO2 supplementation (2 to

5%) may enhance the growth of some of the more fastidiousLegionella spp., such as L.

sainthelensi and L. oakridgensis. This low level of CO2 supplementation will not harm the

growth of L. pneumophila, but CO2 levels higher than 5% may inhibit growth. Since the more

capnophilic species are very rare human isolates, many laboratories do not use

CO2 incubation of media for Legionellaspp.

Plate Inspection

Legionella colonies begin to appear on culture plates on day 3 of incubation. It is very

unusual for the bacterial colonies to appear on plates after 5 days of incubation. Some very

rarely isolated Legionella spp. may require up to 14 days of incubation before growth

appears; this is an extremely rare event. Regardless, it is reasonable to inspect culture

plates on days 1 to 5 and then again at day 14.

The late appearance of Legionella spp. on culture plates can be used to great advantage if a

careful record is kept of the colonies present on days 1 and 2 postincubation. New colonies

appearing after day 2 should be suspected of being Legionella spp. Very

rarely, Legionella spp. may grow from a heavily infected lung (usually from an autopsy of a

fatal untreated case) on day 2, so some latitude in growth rate assumptions needs to be

applied in the case of autopsy lung cultures. Legionella spp. never grow from clinical

specimens on day 1 postincubation, a critical point in the distinction of Pseudomonas

aeruginosa colonies from those of Legionellaspp., as very early colonies of the latter

superficially resemble those of the former.

Proper observation of culture plates requires the use of a dissecting microscope illuminated

with direct light aimed at the plate surface at an approximately 30° angle. Failure to use a

dissecting microscope or use of improper lighting will result in missed positive cultures,

especially when there is mixed bacterial growth on the plates. In addition, very

young Legionella colonies are very small and difficult to see with the naked eye. Therefore,

use of a dissecting microscope can speed up the time to colony detection by as much as a

day.Legionella growth occurs almost exclusively in the first streak quadrant and sometimes

at the edge of the plate.

The size and morphology of Legionella colonies change with time. Very young colonies (day

3) are flat, entire, and 0.5 to 1 mm in diameter; when observed with a dissecting microscope

and incident visible light, they usually have a speckled blue, blue-green, or red color. Within

6 to 24 h of additional incubation, these colonies become smooth, convex, iridescent, and

entire, about 1 to 3 mm in diameter, and look opal-like when observed with a dissecting

microscope (Fig. 2). A thick string may form when a loop is inserted in the colony and then

removed from the colony. In contrast to several mimics, the edges of the colonies are of the

same consistency as the central portion and are not watery and clear. In another 1 to 2

days, the colonies may increase in size up to 5 to 7 mm, become umbonate, sometimes with

tuberculated or inhomogeneous texture and develop spready edges; their iridescent nature

may be lost at this stage. It is these late-stage colonies that are most difficult to distinguish

from non-Legionella spp., making daily plate observation crucial for accurate detection. Very

rarely, some Legionella sp. colonies do not change morphology with prolonged incubation.

Biosafety level 2 precautions should be used for the manipulation of Legionella sp. cultures.

It is safe to inspect culture plates, pick typical colonies, and subculture them on the open

bench in a properly ventilated laboratory. Making an organism emulsion on microscope slides

for the purposes of Gram staining can also be safely carried out on the open bench.

However, vortexing suspensions, sonication, tissue grinding, primary plating, and

manipulations that may result in generation of a high concentration aerosol should be

performed in a biological safety cabinet. No well-documented cases of laboratory-acquired

LD have been reported.

Initial Workup of Suspect Colonies and Look-Alike Bacteria

Colonies suspected of being Legionella spp. should first be stained by Gram stain to ascertain

that the bacteria are small to sometimes filamentous, gram-negative rods. A small amount of

the colony should be emulsified in sterile water or saline on a glass slide. It is important to

completely suspend the bacteria in the liquid, as nondispersed clumps may stain as grampositive

rods. It is also crucial to use 0.1% basic fuchsin counterstain because safranin stains

these bacteria very poorly. Depending on the colony age and on the strain and

species, Legionella spp. taken from plates vary in size from short rods, 0.5 by 5 μm, to very

long, filamentous bacteria, 1 by 25 μm.

Gram-negative bacteria should then be plated to two different media, BYCE α and either

tryptic soy blood (TSB) or BCYEα made without L-cysteine (BCYEα-L), in approximately equal

amounts in a small (1-cm2) area; eight or more isolates can be plated to each plate if needed

(Fig. 3). Rather large amounts of the picked colony should be inoculated to these media to

enable growth after 16 to 18 h of incubation, as the small inocula normally used for other

bacteria may otherwise take several days to produce visible colonies on plates. If only a

small single colony is available, then it can be emulsified in a small amount (~0.5 ml) of

sterile distilled water (not saline) and used for staining, plate inoculation, and

seroidentification. Legionella spp. should grow in 16 to 36 h on BCYEα medium, but not on

TSB agar or BCYEα-L medium; this takes advantage of the L-cysteine growth dependence

of Legionella spp. Sometimes Legionella spp. will grow poorly on TSB or BCYEα-L media, but

at most, only about 10% of the amount of growth on BCYEα will occur. Nutrient carryover

from the primary isolation plate is the explanation for this light growth; this can be proven

by making a subculture of growth on BCYEα-L or TSB on a second plate of the same medium.

Rare Legionella spp. partially lose growth dependence for L-cysteine on serial passage but

even still grow more poorly on BCYEα-L than on BCYEα; these species include L.

spiritensis, L. oakridgensis, and L. jordanis, none of which have been reported to cause more

than two cases of LD each. TSB performs almost as well as does BCYEα-L for determining Lcysteine

dependence for clinical isolates and may be less expensive, depending on the

number of isolates tested per plate. If a blood-containing agar medium is used as the

screening plate, rather than BCYEα-L, for L-cysteine dependence, great care must be taken

that the medium base is not too rich. For example, Brucella blood agar will support L.

pneumophila growth almost as well as BCYEα medium, and other blood-containing media

have been described to do the same (17). The plates are incubated overnight at 35 to 37°C

or until there is visible growth on the BCYEα plate. The relative amount of growth on each

plate is compared to determine if there is L-cysteine dependence. Most Legionella spp.

produce a characteristic dank odor when growing in pure growth that is very specific to the

trained nose.

Common mimics of Legionella spp. colonies on BCYEα plates include Eikenella corrodens, P.

aeruginosa,Flavobacterium spp., and some Bacillus spp. All of these bacteria grow equally

well on BCYEα and BCYEα-Lmedia, but when young, they often grow as speckled colonies on

BCYEα plates. Francisella tularensis can grow well on BCYEα agar but has no resemblance

to Legionella spp. colonies. However, F. tularensis is the only gram-negative bacterium other

than Legionella spp. that exhibits L-cysteine growth dependence. The colony morphology

of F. tularensis is not speckled but rather is opaque and homogeneous. Adding to the

confusion is that some serotyping reagents for Legionella spp. may cross-react with F.

tularensis. There is one case report of the misidentification of F. tularensis as L.

pneumophila (118). When tularemia is suspected, more-stringent safety precautions are

needed. Of note, some Bacillus spp. mimics can stain as gram-negative rods, have

unapparent sporulation, and do not grow on TSB (but will grow on BCYEα-L) (109). With

prolonged incubation, colonies of both E. corrodens and Flavobacterium spp. colonies change

color and no longer resemble Legionellaspp.; E. corrodens colonies become a light to dark

green color and Flavobacterium spp. become a bright yellow color. Very young P.

aeruginosa and Bacillus spp. colonies resemble the speckled flat to slightly convex

youngLegionella spp. colonies but, with prolonged incubation, change their morphology,

making them easily recognizable as non-Legionella sp. colonies. Bordetella pertussis colonies

may appear late on BCYEα plates, and although this bacterium is not cysteine dependent nor

does it possess colony morphology similar to Legionellaspp., B. pertussis has been reported

to be misidentified as Legionella spp., abetted by serological cross-reactivity (90). Because

many different bacteria may cross-react with serological reagents used for typing and

identifying Legionella spp., it is crucial to become familiar with the morphology and growth

characteristics of this genus; relying exclusively on serotyping to identify Legionella spp.

could result in mistaken identification.

Microbiologists should know that some pathogenic fungi and higher bacteria grow well on

BCYEα medium, presenting potential biohazards as well as the opportunity to diagnose

unsuspected infections. Coccidioidesspp. often grow within a day or two on this medium, can

rapidly form arthroconidia, and as such, present a biohazard. Blastomyces dermatitidis also

grows well and converts to the mold phase within a few days. It is likely that other

pathogenic fungi grow equally well on this very rich medium. Nocardia spp. and rapidly

growing mycobacteria often grow quite well on this medium, making BCYEα medium and its

selective variants the plating media of choice for the laboratory diagnosis of nocardiosis

(69, 116). Bacteremia from Nocardia spp. can sometimes be diagnosed by subculture of

blood culture bottles to BCYEα.

IDENTIFICATION Back to top

Basic Identification

Once L-cysteine dependence has been confirmed, further identification of Legionella spp. in a

clinical laboratory relies almost exclusively on serotyping the bacteria, using either

immunofluorescence or agglutination methods (25, 94). Before identification by serotyping is

attempted, the plate should be illuminated with long-wave UV light in a darkroom.

Some Legionella spp. other than L. pneumophila fluoresce a brilliant blue-white color, and

some fluoresce a brilliant red. Such bacteria are best identified by a reference laboratory and

are not L. pneumophila. L. pneumophila fluoresces a very pale yellow-green color, usually

with diffusion of the fluorescent pigment into the culture medium. This is not specific for this

species, and sometimes young cultures are completely nonfluorescent. An excellent and

specific FDA-cleared fluorescein isothiocyanate-labeled monoclonal antibody to all serogroups

of L. pneumophila is available (Monofluo; Bio-Rad). Also, an excellent non-FDA-cleared

fluorescein isothiocyanate-labeled L. pneumophila serogroup 1 (Philadelphia 1 strain)-specific

polyclonal antisera is available (m-TECH, Atlanta, GA [http://www.4m-tech.com/]).

Serotyping can also be performed using non-FDA-cleared latex agglutination antisera,

available in a variety of formats (Oxoid, Basingstoke, United Kingdom; Denka Seiken, Tokyo,

Japan); few data on the performance of these reagents are available.

Since more than 90% of Legionella sp. isolates are L. pneumophila serogroup 1, a reagentand

time-saving technique is to first test with L. pneumophila serogroup 1 antibody; strains

negative with this antibody can then be tested with the species-specific monoclonal antibody.

Extraordinarily rarely, some L. pneumophilaserogroup 1 strains do not react with Philadelphia

1 strain antibodies, making it possible that a serogroup 1 strain could be missed with this

reagent; additional serogroup 1 antibodies are available (m-TECH). Testing has to be

conducted properly for valid results, with the most common errors being false-negative

results because of a prozone phenomenon, and false-positive results from contaminated

buffers, wash solutions, or cross-contamination from controls (25). Strains that fail to react

with L. pneumophila antibodies are best identified by reference or public health laboratories

that have other typing sera and molecular identification methods in use.

Gram-negative rods isolated from clinical specimens that are morphologically consistent

with Legionella spp. and are L-cysteine dependent for growth, can be reported as

presumptive Legionella spp. in the absence of reactivity with typing sera. Gram-negative

rods characteristic of Legionella spp. that react with L. pneumophilamonoclonal antibody

or L. pneumophila serogroup 1 polyvalent antibody can be presumptively identified as L.

pneumophila or L. pneumophila serogroup 1, and the identification can be finalized the next

day once L-cysteine growth dependence is confirmed. If the isolate is nonreactive with L.

pneumophila species-specific antibody or if it is L. pneumophila but not serogroup 1, then

the isolate should be further typed by a reference laboratory. All clinical isolates should be

frozen at −70°C in tryptic soy broth in 10% glycerol and subcultured on a BCYEα slant for

possible analysis by public health authorities.

Advanced Identification

Accurate identification of Legionella spp. other than L. pneumophila and L.

pneumophila serogroup 1 can be quite difficult because of serologic cross-reactivities

between species and serogroups, biochemical inertness, and phenotypic identity of different

species. Identification to the species or serogroup level for these bacteria is usually not of

major clinical significance but may have significant public health and scientific importance.

Infection caused by Legionella spp. other than L. pneumophila or L. longbeachae almost

always occurs in immunocompromised patients, so identification of these

other Legionella spp. could very rarely be a clue to occult immunosuppression (85). LD

caused by Legionella spp. other than L. pneumophila may respond more poorly to

erythromycin therapy (38), but whether knowing the identification of the Legionella spp.

causing infection would influence patient outcome is debatable, especially since erythromycin

is no longer the treatment of choice for severe LD (30). Investigation of LD outbreaks, and

sometimes single cases of the disease, requires knowledge of Legionella spp. identity and

subtype; this identification, if needed, can be performed by a public health or reference

laboratory as long as isolates are frozen. The identification techniques used by reference

laboratories include serotyping using collections of antisera, biochemical characterization,

and sequence-based identification.

Serotyping of Legionella spp. is carried out using polyclonal antisera produced by the U.S.

CDC, other public health agencies, and commercial laboratories (m-TECH, Denka Seiken,

Oxoid, and others). Either immunofluorescence or agglutination reactions are used, with no

clear evidence of superiority of one technique over the other (113). Some producers make

polyvalent serum pools that react with a large number of species or serogroups, which can

be useful to reduce the number of monovalent antisera used. The specificity of the

polyvalent antisera has not been studied in great detail, although at least one product

reacted with a number of non-Legionella spp. (31). This makes it important that monovalent

typing be carried out and that the bacteria meet minimal phenotypic criteria

for Legionella spp. Unfortunately, cross-reactions between different species and serogroups

occur even when using monovalent antisera (1, 95, 99, 106, 111,119). Use of crossadsorbed

polyvalent antisera has been described as a research tool to avoid this problem of

cross-reactions, but these reagents are not available outside some research laboratories

(6, 110, 112). In addition to intragenus cross-reactions, a large number of cross-reactions of

monovalent polyclonal antibodies to non-Legionella spp. have been reported, including P.

aeruginosa, Flavobacterium spp., Bacteroides fragilis,Capnocytophaga ochracea, B.

pertussis, Bordetella bronchiseptica, and possibly Burkholderia

pseudomallei (7, 14,15, 33, 50, 68, 71). Antibody to L. pneumophila serogroup 1 is quite

specific, but otherwise, enough cross-reactions exist to make serological identification only

presumptive. Antisera to newer Legionella spp. are often unavailable.

Monoclonal typing antibodies reduce the number of cross-reactions with gram-negative

bacteria but are only commercially available for the identification of L.

pneumophila (13, 28, 58, 108). However, great care must be used by experienced

microbiologists, as cross-reactions with some non-Legionella spp. have been reported with

the L. pneumophila-specific monoclonal antibody, including S. aureus, yeasts,

and Bacillus spp.

Legionella spp. are relatively inert biochemically and will not be identified using conventional

tube or commercial panel biochemical tests. The few biochemical characteristics described

for Legionella spp. that can be determined in most laboratories, such as oxidase, catalase,

and β-lactamase tests, can be nonspecific and, if performed improperly, falsely negative. Use

of a research biochemical panel has been described to facilitate the identification

of Legionella spp. (115).

Determination of cellular fatty acid and isoprenoid quinone composition by gas-liquid

chromatography and high-pressure liquid chromatography, respectively, can be successfully

used to identify many Legionella spp. (75). These techniques require the use of expensive

equipment, are highly complex, and have been supplanted by molecular methods. The

cellular composition of several newer Legionella spp. have not been studied using these

techniques, are not in commercial databases, or both, making identification of these newer

species by these methods difficult.

The gold standard for the identification of new Legion ella spp. has been DNA-DNA

hybridization analysis (9). This method is tedious, expensive, and labor-intensive and

requires special expertise as well as a large collection of reference DNA standards. As such,

DNA-DNA hybridization analysis is not used to identify already known species.

Molecular identification of Legionella spp. has now replaced other identification techniques in

research and specialty laboratories for several reasons. These include the labor cost and time

required to serotype a strain, serological cross-reactions, limited availability of antibodies to

newer strains, the lack of specific and easy biochemical characterization methods, and the

increasing availability of inexpensive DNA sequencing. Molecular identification methods take

advantage of the specific 16S rRNA or mip gene sequences of the different Legionella spp.

(44, 78, 97, 101). A mip database, procedure instructions, including primer

sequences,mip sequence alignment software, and other genomic software tools are all

available online

(http://www.hpa.org.uk/web/HPAweb&HPAwebStandard/HPAweb_C/1195733805138).

Sequence analysis of 16S rRNA has been used to identify several new Legionella spp.

(43, 76, 80, 93). Partial sequencing of 16S rRNA (500 bp) is able to accurately identify

all Legionella spp. to the genus level and all L. pneumophila and about 90%

of Legionella spp. other than L. pneumophila to the species level; incorrect species-level

identification is a problem for the more unusual Legionella spp. (120). Sequence of

the rpoB gene has been shown to distinguish between species as well as or better than 16S

rRNA or mip gene sequencing (72). Intergenic 16S-23S ribosomal spacer PCR analysis has

been successful for species identification and does not require DNA sequencing (100). The

relative performances of mip and 16S rRNA sequencing is not known, but partial 16S rRNA

sequencing should not solely be relied upon when an unusual Legionella sp. is identified.

Limited numbers of reference strains, especially of some unusual species, have been tested

by these methods, leaving open the possibility of incorrect classifications. A commercial plate

DNA hybridization assay has been reported to correctly identify 23 different Legionella spp.

(36) and is distributed in Japan (Kobayashi Pharmaceutical, Osaka, Japan).

Molecular methods cannot be used to accurately serogroup L. pneumophila, as there are

several reports of genotypic discordances for identical serogroups and vice versa

(10, 56, 81). Sequencing of the L. pneumophiladnaJ gene appeared to be able to distinguish

between some, but not all, L. pneumophila serogroups (78); different serogroups that had

very similar genotypes had been shown by other methods to have discordant genotypes and

serogroups. More extensive testing of this method needs to be performed before it is put into

routine use.

TYPING SYSTEMS Back to top

Typing of Legionella spp. is important for public health investigations to help link culturepositive

environmental sites with clinical isolates during an epidemic of the disease. Typing

cannot be used by itself to determine the environmental source of an outbreak and must be

accompanied by an epidemiologic investigation. Otherwise, incorrect conclusions may be

made about epidemic sources (57, 73). This problem is due to clonal distributions of

environmental and clinical Legionella spp. (3, 65, 77, 96) and to the poor specificity of some

typing techniques (22, 96).

Monoclonal antibody typing plays a major role in subtyping L. pneumophila serogroup 1

isolates and, when used with molecular methods, can increase typing specificity (42, 45, 96).

Used by itself, monoclonal antibody typing may not be specific enough to distinguish

between closely related strains.

Sequence-based typing appears to be the most specific and precise molecular subtyping

system for both L. pneumophila and L. pneumophila serogroup 1 (98). A standardized

pulsed-field gel electrophoresis method yields reproducible results and is used as a reference

typing method by one national laboratory (3).

SEROLOGIC TESTS Back to top

LD can be diagnosed by demonstration of an increase in antibodies to killed bacterial cells

(26). Indirect immunofluorescent assay is considered the gold standard method. While most

patients develop both immunoglobulin G (IgG) and IgM responses, some develop only IgMonly,

IgG-only, or IgA-only responses, making it necessary to test for total immunoglobulin

response and not just IgG. In addition, IgM antibodies may persist for as long as a year after

infection, making IgM presence a poor marker of acute disease (88). About 75% of patients

with culture-proven nosocomial L. pneumophila serogroup 1 LD develop seroconversion to

the bacterium, whereas the test seems to have higher sensitivity in LD epidemics.

Seroconversion requires weeks to months after infection, with only about a 50%

seroconversion rate after 2 weeks; for optimal test sensitivity, acute-phase serum should be

frozen and convalescent-phase sera should be collected at 2, 4, 6, 9, and 12 weeks

postinfection. Parallel testing of sera is required for the best specificity. The most specific

testing is for seroconversion to L. pneumophila serogroup 1 only and the least specific is the

use of polyvalent antigen preparations, with approximate test specificities of 99 and 90 to

95%, respectively. Serologic diagnosis is best used for epidemiologic studies because of the

retrospective nature of serologic diagnosis and limitations on test specificity and sensitivity.

ANTIMICROBIAL SUSCEPTIBILITIES AND

SUSCEPTIBILITY TESTING Back to top

The antimicrobial susceptibility of L. pneumophila grown in broth or on agar can give results

that have no clinical correlate. This is because of the intracellular location of the bacterium in

human infection, to which not all antimicrobial agents gain access and retain activity (23). In

addition, the complex broth and agar media used to grow L. pneumophila inactivate many

drugs. There is no indication for performing antimicrobial susceptibility testing

against Legionella spp. except in a research setting, when correlative studies of intracellular

and experimental animal infection models can be performed. The microbiologist must not

assume that a particular drug will be effective for the treatment of LD simply because the

drug is active against L. pneumophila in vitro, nor should there be an assumption that drugs

having low MICs against the bacterium in vitro will be more clinically effective than will drugs

with higher MICs for the organism. Antimicrobial resistance to drugs used for LD treatment

has never been documented to be responsible for clinical treatment failures and has only

been demonstrated under laboratory conditions.

Antimicrobial agents that have good intracellular activity against L. pneumophila include

most macrolide, tetracycline, ketolide, and quinolone antimicrobial agents (23, 30). No β-

lactam agent or aminoglycoside has acceptable intracellular activity against the bacterium. It

is unknown if the intracellular activity of antimicrobial agents against L. pneumophila can be

extrapolated to all other Legionella spp. and to treatment of infections caused

by Legionella spp. other than L. pneumophila. Some of these other Legionella spp. may

reside in different subcellular compartments than does L. pneumophila and, thus, may

respond differently to antimicrobial agents (91). However, macrolide and quinolone

antimicrobials appear to be effective for the treatment of LD caused by L. micdadei, L.

longbeachae, L. bozemanae, and L. dumoffii (85).

EVALUATION, INTERPRETATION, AND REPORTING OF

RESULTS Back to top

Multiple laboratory methods have to be used for optimal laboratory diagnosis of LD. Culture

of Legionellabacteria from sputum, lung, or other respiratory sites is the most specific

(100%) method for diagnosis of the disease, very sensitive (~80 to 90%) in severe

untreated disease, and insensitive (~20%) in those with mild disease. Culture may be the

only positive diagnostic test, especially when Legionella bacteria other than L.

pneumophila are causing the infection. In addition, a culture isolate can be used to sort out

the source of an epidemic, unlike any of the other diagnostic tests. Because of the technical

difficulty of culture diagnosis, its expense, and its low sensitivity for nonsevere disease,

several alternative diagnostic methods have been developed. The antigenuria assay is more

sensitive than culture for the detection of community-acquired disease, especially epidemic

disease. Still, the antigenuria test is only about 60% sensitive in the best of circumstances,

performs poorly for detection of nosocomial infection, and detects almost exclusively L.

pneumophila serogroup 1. Antibody detection complements other laboratory diagnostic

methods but is retrospective, as it requires seroconversion for greater test specificity and

sensitivity. In addition, antibody results obtained using only polyvalent antigens must be

viewed with circumspection. Commercial molecular amplification tests are just being

marketed and have yet to be thoroughly evaluated; prior to adoption, more extensive

studies are required. The performance of all laboratory diagnostic tests for non-L.

pneumophilaserogroup 1 LD is unknown but presumably is not as good as it is for the

diagnosis of L. pneumophilaserogroup 1 disease.

A major hindrance to the evaluation of all laboratory diagnostic methods is the lack of a good

gold standard for diagnosis of LD. Culture diagnosis, while very specific, is known to be

imperfect and of limited sensitivity, especially for epidemic LD. Relative performance of the

diagnostic tests is often compared to culture diagnosis, which tends to overestimate test

sensitivity and underestimate specificity.

Positive cultures for all Legionella spp. are virtually diagnostic of LD, providing there are

supportive clinical findings such as pneumonia. In contrast, single serum specimens showing

elevated antibodies to Legionellaspp. or to L. pneumophila are often not the result of LD.

Only rises in antibody titers to L. pneumophilaserogroup 1, a test not usually commercially

available, are specific enough for diagnosis, but even then, use of appropriate techniques is

required for optimal specificity. Detection of L. pneumophila serogroup 1 antigenuria is

almost as specific as a positive culture once heat-labile factors capable of causing falsepositive

tests are excluded. Positive test results must be reported promptly to the patient’s

clinician as well as to infection control and public health authorities.