Consensus Date:4 April 2007
1 PHLN Summary Laboratory Definition
1.1.1 Definitive Criteria
- Isolation of Legionella species; or
- Seroconversion or significant increase in serum antibody level; or
- Detection of Legionella antigen in urine.
1.1.2 Suggestive Criteria
- Single high antibody level to L. pneumophila or L. longbeachae ; or
- Detection of Legionella antigen in specimens by Direct Fluorescent Antigen (DFA); or
- Detection of Legionella specific target by Polymerase Chain Reaction (PCR).
Legionellosis is an acute infection—usually in adults and often associated with underlying disease—due to bacteria belonging to the genus Legionella. It usually presents with pneumonia but systemic features are common and may be predominant. Wound infection occasionally occurs. Legionella pneumophila serogroup 1 (Lp1) is the most common cause of legionellosis. However, in Australia, up to half of all cases are due to L. longbeachae serogroup 1. The source of infection is environmental, although in sporadic cases this is rarely identified. Outbreaks of legionellosis (usually due to Lp1) occur quite commonly and often can be traced to a common source, such as an air-conditioner cooling tower. Cases of infection with L. longbeachae serogroup 1 have been associated with commercial potting soils. Infections due to other serogroups and species are less common, usually sporadic and more likely to occur in individuals with underlying disease associated with immune deficiency or respiratory pathology. The diagnosis of legionellosis is made by a) isolation of a Legionella species from a clinical specimen—usually respiratory, occasionally blood or wound swab; b) a significant change in the level of serum antibody (or seroconversion) against a Legionella species, using a suitably validated test; c) detection of specific urinary antigen (Lp1 only) or d) specific Legionella nucleic acid in an appropriate specimen.Top of page
Legionellae are fastidious organisms requiring cysteine and other essential growth promoting factors for their successful isolation from clinical material. Legionella grows aerobically but requires high humidity.
3.1.1 Suitable Specimens
Bronchial washings and bronchoalveolar lavages are the specimens of choice and, if possible, should be collected before antibiotic therapy is commenced; pleural aspirates, lung or other tissue are suitable if available. Use of saline during collection of specimens should be avoided as this may inhibit the growth of Legionella species. Where this is unavoidable as with bronchoscopy specimens, the specimen should be cultured as soon as possible. If delays are likely to occur the specimen may be centrifuged and resuspended in a broth such as TSB.
Expectorated sputum and tracheal aspirates are less satisfactory as they are likely to be heavily contaminated with oral flora and contain relatively few legionellae. They should be cultured to non- selective and selective media and additional plates inoculated after treatment with acid or heat to reduce the growth of contaminants.
Most laboratories use buffered charcoal yeast extract agar with added ferric pyrophosphate, L -cysteine and alpha-ketoglutarate (BCYE). Selective agar commonly used is BCYE containing vancomycin, polymyxin B and pimafucin (BCYE VPP),or cefamandole, polymyxin B and anisomycin (BMPA). Some laboratories use modified Wadowsky and Yee medium MWY which contains anisomycin, polymyxin B and vancomycin .
Legionellae usually require 48 hours incubation before growth is visible and colonies may not be visible for up to 5 days or more. Culture plates should be incubated and examined for typical colonies using a dissecting microscope with side lighting daily for up to 10 days incubation. Such viewing highlights the iridescence of early growth and shows the ground glass appearance of mature colonies.
3.1.3 Suitable sensitivity
Depends on the quality of the specimen. The presence of contaminants may inhibit the growth of Legionella, particularly when the specimen contains low numbers. Antibiotic treatment may inhibit growth.
3.1.4 Test specificity
The isolation of any Legionella species from a clinical specimen is considered to be significant.
3.1.5 Predictive values
A negative culture does not exclude the diagnosis of legionellosis.
3.1.6 Suitable test acceptance criteria
Pale staining gram negative rod, often pleomorphic, which grows only on BCYE and not on Blood Agar or BCYE without cysteine.
3.1.7 Suitable internal controls
Need to ensure media will support growth of legionellae in a competitive environment by inoculating samples of each batch with L. pneumophila and L. longbeachae.Top of page
3.1.8 Suitable test validation criteria
Isolation of Legionella species is the gold standard.
3.1.9 Suitable external QC program
3.1.10 Special consideration
L. oakridgensis requires cysteine only for primary isolation and subcultures will grow without cysteine. Francisella tularensis has similar growth characteristics but is a very uncommon isolate in Australia.
3.2 Identification of Legionella species
Definitive identification of Legionella species is based on DNA/DNA hybridisation, gas chromatography (GC) for long chain fatty acids and high performance liquid chromatography (HPLC) for ubiquinones. Recently speciation based on sequencing of the mip gene has been developed as a rapid method of confirming the identification of known species although the full description of a new species still requires the extended range of definitive tests.9 A database for mip sequences may be accessed at http://www.hpa.org.uk/Topics/InfectiousDiseases/InfectionsAZ/LegionnairesDisease/
In routine laboratory practice serotyping is used to identify presumptive L. pneumophila and L. longbeachae but is unreliable for other species owing to the high degree of cross-reactivity among them. This assay is performed by reacting whole bacterial cells with antibody raised in rabbits to known Legionella species. The reaction can be performed on a slide using sera bound to latex particles to visualise the agglutination or by UV microscopy using fluorescently labelled antisera (DFA). While serological identification by latex agglutination or DFA is acceptable for most sporadic clinical isolates of
L. pneumophila and L. longbeachae, all isolates should be kept and/or sent to a reference laboratory and fully identified if a) there is a suspected outbreak or b) the isolate is unusual or c) the diagnosis of legionellosis is uncertain or there is a possible medicolegal issue.Top of page
3.3 Antigen Detection Tests
3.3.1 Direct fluorescent antigen (DFA)
DFA may be used directly on specimens including paraffin sections after dewaxing. However it is rarely used except during outbreaks or when unfixed tissue is not available because of the laborious nature of the test, its lack of sensitivity and the low rate of positives seen routinely. DFA will not be considered further.
3.3.2 Urinary antigen test 3,5,6
Antigen to L. pneumophila serogroup 1 in urine may be detected by immunoassay and rapid format immunochromatographic kits. In Australia the most commonly used test is the Binax NOW ICT for the qualitative detection of L. pneumophila serogroup 1 antigen in urine. Reference laboratories may also perform the Binax Legionella Urinary Antigen Enzyme Immunoassay, which employs microtitre trays coated with polyclonal rabbit antibody specific for a heat stable antigen believed to be a lipopolysaccharide and as such can be concentrated by ultrafiltration of the urine to enhance low or borderline results. The ICT is faster than the EIA format (15 vs 90 minutes) and more suitable for use with smaller numbers of specimens.
18.104.22.168 Suitable specimen
22.214.171.124 Test sensitivity
High (85% to 94%) 5 in community acquired, 50% in nosocomial infections depending on time of collection relative to onset of disease – see below. The Binax EIA is slightly more sensitive than the Binax NOW ICT test. Whilst the Binax NOW ICT has good specificity and sensitivity for Legionella infection, occasional borderline or false negative results can occur. If in doubt of any ICT result, laboratories may wish to refer the urine to a laboratory which also performs the Binax EIA. Alternatively, ask for a repeat urine as antigen excretion can be intermittent and may persist for months.
126.96.36.199 Test specificity
Close to 100% to Legionella spp.
188.8.131.52 Predictive values
Positive—~ 100% for Legionella spp, negative—85% for community acquired L. pneumophila serogroup 1 infections.
184.108.40.206 Suitable test acceptance criteria
Refer to the kit instructions.
220.127.116.11 Suitable internal controls
Positive and negative controls supplied in the kit; or human urine containing L. pneumophila serogroup 1 antigen (positive control); normal human urine (negative control).
18.104.22.168 Suitable validation criteria
Defined by the kit manufacturer.
22.214.171.124 Suitable external QC program
None as yet available in Australia; programme available in Europe through the European Legionnaires' Disease Surveillance Network (ELDSNet) - formerly European Working Group on Legionella infections (EWGLI) http://www.ewgli.org/
126.96.36.199 Special considerations
An antigen peak is found at 5–10 days after onset of symptoms and usually diminishes rapidly to barely detectable 21 days after onset of symptoms. In rare cases antigen may continue to be excreted for months after infection has resolved.6 These kits are very sensitive for detecting most subtypes isolated from community acquired L. pneumophila 1 infections, which are usually caused by Pontiac strains. However, they are less sensitive when infections are caused by non Pontiac strains, particularly the Bellingham subtype. This may have implications for their use in hospital settings.6 These tests may also detect other serogroups of L. pneumophila and other Legionella spp. Therefore positive urinary antigen results should always be verified by culture or PCR if possible. In addition, if legionellosis is strongly suspected on clinical grounds, culture or PCR should be performed since a negative urinary antigen test cannot exclude infection due to other Legionella Species.Top of page
3.4 Nucleic Acid (NA) Based Tests
Laboratories have been slow to introduce PCR assays for Legionella detection, owing in part to the failure of clinicians to submit respiratory specimens from suspect cases, relying, instead, on urinary antigen detection or serology for diagnosis. This has changed recently with the development of highly sensitive real-time PCR assays which are capable of detecting Legionella in serum and urine.1,6,8 PCR has the advantage over urinary antigen assays of detecting Legionella other than L. pneumophila serogroup 1. Most in-house assays described in the literature target the genes for rRNA or mip (macrophage infectivity potentiator). Becton Dickenson Biosciences’ BD ProbeTec ET L. pneumophila (LP) Amplified DNA Assay has been approved by the US Food and Drug Administration for sputum specimens but it is not marketed in Australia. EnviroAMP, a kit marketed by Perkin Elmer for environmental samples has been available since the early 1990’s but is not licensed for use on human specimens.
3.4.1 Suitable Specimens
Respiratory tract specimens, urine, serum.
3.4.2 Test sensitivity
Measured in patients who fit the laboratory case definition (culture positive or urinary antigen positive or seroconversion or significant increase in antibody level).
5S rRNA: 54.4%–80% in serum, 80% in respiratory specimens, 46%–86% in urine specimens,8
16S rRNA: 30.9% in serum
mip: 52.9% in serum
Sensitivity highest early in disease and assay may be positive up to one month post disease onset. PCR in serum may be positive for longer than the urinary antigen test.
3.4.3 Test specificity
Variable, depending on primer design but approaches 100% in some studies.
3.4.4 Predictive values
Positive predictive value is high if primer and assay design is optimal. Negative predictive value depends on the site of specimen collection and time post onset of disease.
3.4.5 Suitable test acceptance criteria
Results for control samples obtained as expected.
3.4.6 Suitable internal controls
As recommended in NPAAC guidelines: Laboratory Accreditation Standards and Guidelines for Nucleic Acid Detection and Analysis . Controls should be designed to detect sample inhibitory activity and external contamination by environmental Legionella.
3.4.7 Suitable validation criteria
Consistent with NPAAC Guidelines:
3.4.8 Suitable external QC program
3.4.9 Special considerations
PCR can be used to detect Legionella in tissue in paraffin blocks after dewaxing and may be useful for diagnosis when unfixed specimens are not available. Since Legionella bacteria are ubiquitous in the environment caution should be exercised in using these tests to confirm cases of legionellosis if there is any likelihood of environmental contamination of the specimen. Some commercial DNA extraction kits have been contaminated with Legionella DNA, producing false positives when used for DNA extraction of specimens.8Top of page
3.5 Serological Tests
Serum antibodies produced in response to Legionella infection can be measured by a variety of methods including the indirect immunofluorescent assay (IFA) and EIA. In Australia the vast majority of laboratories use either in-house or commercial IFA assays. A commercial, automated CFT for L. pneumophila serogroups 1-6 is also available in Australia.
For the IFA assay, dilutions of patient sera are reacted with whole killed Legionella cells fixed to microscope slides. Antihuman immunoglobulin conjugated with fluorescein isothiocyanate is then added. If Legionella antibody is present in the serum the bacteria will fluoresce when the slides are viewed by UV microscopy.2
There are at least two commercial EIA assays for Legionella available overseas although none is presently in use in Australia. A recent publication 7 and local surveys have shown that commercial assays are generally inferior to in house EIA assays. EIA assays are generally designed to provide a sensitive screen for legionellosis and detect IgM using L. pneumophila 1 or L. longbeachae 1 sonicated whole cells as antigen. Since IgM to Legionella spp. can remain elevated for years and numerous cross-reactions occur among the non- pneumophila legionellae and other organisms, a single positive IgM level should be regarded with caution in sporadic cases.
3.5.1 Suitable specimens
Ideally, paired sera collected as soon as possible after the onset of illness and 3-6 weeks later. There may be up to 9 weeks delay before seroconversion can be detected.
3.5.2 Test sensitivity
Seroconversion is defined as a four-fold increase in titre of IF antibody against heat killed L. pneumophila serogroup 1, to ≥ 1:128. Sensitivity is 70–80% overall, and up to 90% if collection of convalescent serum is delayed to 6 weeks after onset.
Similar sensitivity has been found for antibody seroconversion to the heat killed L. longbeachae serogroup 1 antigen. Four fold rises to ≥ 1:512 were detected in 11/12 patients with culture-confirmed L. longbeachae infection.6
A single high titre of 1:512 or higher in either L. longbeachae or L. pneumophila is a sensitive indicator of infection with Legionella but may represent past infection or rarely infection with another species (see below).
In published reports commercial EIA sensitivities vary enormously from less than 20% when compared with IFA in randomly selected sera 7 to ~70% when tested in patients from a single outbreak.10 EIA assays developed in house using high quality antigen have been shown to have much higher sensitivities (82%7 ) than commercial kits. IgM measured by EIA can become positive earlier in the course of illness than IFA.Top of page
3.5.3 Test specificity
Specificity of IFA tests, using L. pneumophila serogroup 1 antigens, is reported to range from 95–99.9%.6
Specificity of IFA for other Legionella species is variable and results should be interpreted with caution because of the numerous cross reactions which occur among Legionella and other bacteria. For example, cross reactions in IFA tests using non-L. pneumophila serogroup 1 antigens have been reportedin patients with pneumonia or bacteraemia caused by Pseudomonas, Haemophilus, Mycobacteria, Bordetella, Chlamydia, Rickettsia, Campylobacter and Bacteroides species, Enterobacteriaceae and Coxiella burneti.2
IgM EIA based tests have been reported to be slightly less specific (97%) than IFA when acute and convalescent sera are tested for L pneumophila 1.10 There are no published data for specificity of IgM EIA for L. longbeachae, but, as with IFA, cross-reactions among legionellae and other organisms are highly likely to affect specificity. Therefore, single positive results should be interpreted with caution.
3.5.4 Predictive values
Positive tests: four-fold rise in antibody titre is highly predictive of legionellosis (L. pneumophila to 1:128 or higher and L. longbeachae to 1:256 or higher). Serological tests for Legionella species other than L. pneumophila serogroup 1 are predictive of legionellosis caused by the Legionella species antigen used or a related Legionella species.
Negative tests: In one study approximately 10% of patients with culture proven L. longbeachae failed to seroconvert after 3 months10 and up to 20% may not seroconvert to L. pneumophila 1 antigen in culture proven cases.12
A single high antibody titre of ~512 or greater against L. pneumophila is rarely seen in control subjects and, if detected in a patient with suspected legionellosis, itmay be significant and should be notified.
Similarly, single high antibody titres against L. longbeachae of ~512 or greater are rare (1–2 %) in healthy controls and may be significant in a patient with compatible illness. Actual cut-off titres vary in different laboratories, based on local evaluation of the test using statistically relevant samples of control sera.
Positive tests: Single high levels to L. pneumophila 1 are highly predictive of disease but may represent past infection.
Single high levels to L. longbeachae may represent past infection or crossreactions. The significance of positive results should be determined using another serological test or diagnostic method such as urinary antigen assays.
Negative tests: predictive values of commercial kits may be low, but much higher in well designed and validated in- house assays.
3.5.5 Suitable test acceptance/validation criteria
Consistent with NPAAC Guidelines:
Commercial kits: according to manufacturer’s instructions.
IFA: Positive control/s should be within one doubling dilution of the established endpoint.
3.5.6 Suitable internal controls
IFA: Positive control for each antigen tested should be titrated to at least two dilutions past the established endpoint.
3.5.7 Suitable external QC program
RCPA QAP P/L Serology and Microbiology QAPs
3.5.8 Special considerations
The endpoint titre for IFA is the reciprocal of the highest serum dilution giving 1+ fluorescence in at least 80% of the organisms. A suitable means of determining the percentage of organisms fluorescing is to use a fluorescent microscope equipped with a dark field condenser. The practice of running IFA tests using pooled antigens, containing multiple serogroups of L. pneumophila may give misleading results owing to the variability of fluorescent staining often seen with this species. The use of a fluorescence-labelled conjugate that detects all classes of antibody is recommended for optimum sensitivity.IgM antibody can persist for a long time and is not reliable for separating new from old disease.
All in house assays should be properly validated in an adequate number of culture confirmed cases. Laboratories should establish the reactivity to L. pneumophila and L. longbeachae of sera from healthy controls in their region and if possible from patients with bacteriologically confirmed pneumonia not caused by Legionella species. These studies may then be used to identify patients whose antibody levels are suggestive of infection and require further action including notification to the local Public Health Unit.
It is recommended that, to avoid unnecessary testing and potentially false negative results, acute serum specimens collected within 2 weeks of onset of illness should be stored until convalescent sera are received and then tested in parallel. An exception should be made for cases in which the first serum is collected relatively late in the illness.Top of page
Several methods are useful for subtyping Legionella species including pulsed field gel electrophoresis (PFGE), restriction fragment length polymorphisms (RFLP) using Southern hybridisation, random amplified polymorphic DNA (RAPD), AFLP, multiple enzyme electrophoresis (MEE), MLST and consensus sequence based typing (SBT). The choice of method depends on the preference of the laboratory performing the test. The UK Health Protection Agency hosts a website for Legionella identification http://www.ewgli.org with a link to the European Working Group for Legionella Infections (EWGLI) which hosts databases for AFLP and consensus sequence based typing (SBT) profiles to provide standardised subtyping schemes throughout the EU.
Subtyping is most helpful in demonstrating differences between isolates. Apparent identity of isolates does not prove that they have originated from the same source but may be helpful to support epidemiological or clinical data.
A panel of seven monoclonal antibodies for subtyping L. pneumophila 1 was developed in 1987, by a collaboration among English, US, Canadian and French laboratories and is used extensively overseas.4 The full panel distinguishes ten main subtypes of L. pneumophila 1: the Pontiac subtypes , Philadelphia, Knoxville, Benidorm, Allentown, France, and the non- Pontiac subtypes, Olda, Heysham, Camperdown, Oxford and Bellingham. Names were given according to where the first isolate of each subtype was isolated.
Only three monoclonal antibodies from this panel are available in Australia. They were developed at CDC by McKinney and are available from the IMVS by request. Monoclonal antibody 2 is particularly useful for identifying Pontiac strains, which have caused all known community outbreaks of L. pneumophila serogroup 1 and most sporadic cases in Australia. Non-Pontiac strains do not react with Mab 2 but may be associated with nosocomial infections. Mab 2 can be used to identify Pontiac strains and therefore provide evidence for the presence of pathogenic subtypes in environmental samples from sources implicated in community outbreaks where there is not a human isolate for comparison.
4.2 Special considerations
Subtyping, to identify a possible environmental source of legionellosis caused by any Legionella species, should only be performed when there is clear epidemiological evidence to implicate that source. Most jurisdictions now have more than a decades worth of profiles obtained by subtyping of L. pneumophila 1 isolates. Comparison of profiles demonstrates that each jurisdiction has its own distinct single or multiple strain/s causing the majority of sporadic and outbreak associated infections . For example, a single subtype has been isolated from nearly half the sporadic cases in South Australia and associated with both outbreaks there in widely different localities (Kangaroo Island and Adelaide). Pontiac strains have never been isolated from an environmental source in SA except when implicated with a case but this might not be true of other States.
There is very good circumstantial evidence that commercial potting soils are a source of infection with L. longbeachae serogroup 1, and potting soils found in the homes of people with legionellosis have been shown, by RFLP and PFGE subtyping, to contain strains similar to those isolated from patients.11 However, this does not necessarily prove that potting mix from a particular commercial supplier is the source of infection.Top of page
1. Diederen BMW et al. (2007). Evaluation of real-time PCR for the early detection of Legionella pneumophila DNA in serum samples. J Med Microbial 56: 94-101.
2. Edelstein, PH, (1997) Manual of Clinical Laboratory Immunology. 5th ed , Rose NR et al., (Ed). American Society for Microbiology, Washington, DC.
3. Helbig JH et al. (2001) Detection of Legionella pneumophila antigen in urine samples by the Binax NOW immunochromatographic assay and comparison with both Binax Legionella Urinary Enzyme Immunoassay (EIA) and Biotest Legionella URIN Antigen EIA. J. Med. Microbial. 50:509-516.
4. Joly JR et al. (1986). Development of a standardized subgrouping scheme for Legionella pneumophila serogroup 1 using monoclonal antibodies. J Clin Microbiol 23:768-771
5. Kazandjian et al. (1997) Rapid diagnosis of Legionella pneumophila serogroup 1 infection using enzyme immunoassay for detection of urinary antigen. J Clin Microbiol; 35: 954-956.
6. Lindsay DSJ et al. (2004) Laboratory diagnosis of legionnaires’ disease due to Legionella pneumophila serogroup 1: comparison of phenotypic and genotypic methods. J. Med. Microbiol. 53:183-187
7. Malan AK et al. (2003) Comparison of two commercial enzyme-linked immunosorbent assays with an immunofluorescence assay for detection of Legionella pneumophila types 1 to 6. J Clin Microbiol. 41:3060-3063.
8. Murdoch DR (2004). Molecular genetic methods in the diagnosis of lower respiratory tract infections. APMIS 112: 713-727.
9. Ratcliff et al (1998)Sequence-based classification scheme for the Genus Legionella targeting the mip gene. J Clin Microbiol. 36:1560-1567.
10. Rojas A et al. (2005). Value of serological testing for diagnosis of legionellosis in outbreak patients. J Clin Microbiol 43:4022-4025.
11. Steele, TW. (1997) Control of Legionella spp particularly Legionella longbeachae in potting mixes. Horticultural Research & Development Corporation . Report no. RDC NT234.
12. Winslow, WE, Steele, TW. (1993). Indirect immunofluorescent antibody tests with Legionella longbeachae serogroup 1 antigen in confirmed infections. In Legionella: Current Status and Emerging Perspectives. Barbaree et al Ed. American Society for Microbiology, Washington, DC.
13. Wilkinson, HW. (1987) Hospital-Laboratory Diagnosis of Legionella Infections. CDC, Atlanta
14. Winn WC JR, (1999) Manual of Clinical Microbiology, 7th ed. Murray PR, et al.(Ed), American Society for Microbiology, Washington DC.
Further references concerning assay characteristics for commercial kits may be found in the package inserts.