Elisa Legionella

Legionella pneumophila antibody assays
An evaluation of commercial kits and reagents for the

estimation of Legionella pneumophila antibody levels.
Respiratory and Systemic Infection Laboratory, CPHL Central Public

Health Laboratory, Colindale
Spring 1999
TG Harrison
N Doshi
G Hallas
RC George
Contents
Synopsis of the study 3

Background 6
Legionella pneumophila assays evaluated 8
Methods 15
Specimen panel 16
Results 19
Conclusions 37
Acknowledgements 37
References 38
Technical appraisal 39
Appendices
Outline evaluation protocol sent to UK distributors 43
The in-house protocol used for the MRL slides 45
Comparison of results obtained with the RSIL IFAT,
Using FYSA, and each commercial kit 47
Addresses of participating UK distributors 49
RSIL Evaluation Report Legionella pneumophila antibody assays 2

Synopsis of the study
BACKGROUND
In England and Wales serological confirmation of suspected

Legionnaires’ disease has traditionally relied upon reagents supplied
by the PHLS. However following the decision of the PHLS Board that
from July 1994 reagent production at CPHL would cease, stocks of
Legionella pneumophila antigens have become depleted: L.pneumophila
formalin-killed yolk-sac antigen (FYSA) is no longer available and
supply of rapid microagglutination (RMAT) antigen will cease at the
end of July 1999. To allow laboratories in England and Wales to
make an informed choice when selecting commercially available
replacements, RSIL undertook an evaluation of all suitable
commercial alternatives.
MATERIALS AND METHODS
The following kits and reagents were evaluated. EIA kits: Sigma
Legionella IgG/A/M and Zeus L. pneumophila Sgp 1-6. IFAT kits: Zeus
L. pneumophila Sgp1, Zeus L. pneumophila Sgp 1-6, Gull Laboratories
L. pneumophila 1-6. IFAT antigen slides: MRL L. pneumophila/
Legionella spp. heat-killed and formalin-killed antigen slides.
Few, or no, data are available for these commercial kits which
specifically relate to the situation in England and Wales.
Consequently the first part of the study was to determine the
specificity of each assay using 500 blood donor sera from apparently
healthy adults. The specificity of each kit/reagent, in the context of
the differential diagnosis of pneumonia requiring admission to
hospital, was then determined using sera from 114 patients with
microbiologically confirmed non-legionella respiratory infections
(S. pneumoniae, M. pneumoniae, Chlamydia spp. etc.). A small number
of sera giving falsely positive results due to ‘campylobacter’ antibodies
were also examined. Assay sensitivity was determined using sera
from 111 cases of proven L. pneumophila serogroup 1 infection and six
culture confirmed cases of L. pneumophila non-serogroup 1 infection.
Each of these parameters was determined in two ways: firstly for ‘any
positive result’ irrespective of whether or not the result would be
considered diagnostic of L. pneumophila infection. This was a
specimen based analysis and was intended to give an indication, for
each assay, of the likely significance that the testing laboratory should
place on any positive result it might detect. The second approach was
to determine the specificity and sensitivity for each kit with regard to a
‘diagnostic result’. This was a patient based analysis that took into

account all sera for a particular patient. Positive and negative
predictive values were calculated. A total of 932 sera were examined
in all assays.
RESULTS
Specificity/seroprevalence in a healthy population - Specificties

ranged from 80.8%, for the Zeus Sgp 1-6 IFAT antigen, to 98.8% for
the MRL formalin Sgp 1 antigen. The Zeus definition for the endpoint
was found to be at variance with common practice. If the definition for
‘2+’ intensity was taken as the endpoint then the specificity was much
better being 97.4% for the polyvalent and 99.8% for the monovalent
IFAT antigens respectively.
Specificity - The 219 sera from the 114 non-legionella pneumonia

patients were all examined by each assay. The specificities for ‘any
positive result’ ranged from 89.5% for the Gull 1-6 IFAT, to 99.1% for
Zeus 1 IFAT. Both EIAs gave specificities of 98.2%. The specificities
for a ‘diagnostic result’ ranged from 89.5% for the Gull 1-6 to 100% for
the MRL formalin IFAT antigens.
Sensitivity - Estimates of sensitivity were made using sera obtained

from 111 patients with proven L. pneumophila serogroup 1 infection.
The sensitivity for ‘any positive result’ ranged from 69.7% for the MRL
heat-killed IFAT antigen to 95.1% for the Gull 1-6 IFAT antigen.
Sensitivities for a ‘diagnostic result’ ranged from 63.0% for the MRL
IFAT antigens to 97.8% for the Gull 1-6 IFAT. The high sensitivity of
the Gull 1-6 antigen is not unexpected given its correspondingly low
specificity.
Estimates of the Positive and Negative Predictive Values - PPV

and NPVs were calculated for a ‘diagnostic result’ assuming a
prevalence of L. pneumophila infection of either 2% or 5% of
pneumonia patients requiring admission to hospital. For the EIAs the
PPV of a ‘diagnostic result’ was between 48% and 73.2%, for the Sgp1
heat-killed antigens (MRL and Zeus) it was between 58.8% and 80.5%,
and for the MRL formalin Sgp 1 antigen it is estimated to be 100%.
However for the polyvalent IFAT kits it ranged from as low as 16.0%
for the Gull 1-6 to 60.3% for the Zeus 1-6.
Campylobacter cross reactions – It was found that all the kits gave
positive results with sera from some of the ‘campylobacter’ patients
and all gave diagnostic titres with sera from at least one of the
patients. This confirms previous work that the cross-reaction is not
dependent upon the type of antigen or assay format used.

Sequential testing - The possibility of using an EIA as a screening
test followed by one of the IFATs as a confirmatory test, for any
samples found to be EIA positive, was examined. The sensitivity,
specificity, PPV and NPV were calculated for such sequential testing. It
was clear that this approach gives very high PPVs (~100%) but that
the sensitivity for each combination was a little lower than single
assay testing.
CONCLUSIONS
The simplest option for the replacement of the CPHL L.pneumophila

serogroup 1 FYSA is to substitute for it the MRL formalin-killed IFAT
antigen. Results obtained for the L. pneumophila serogroup 1
component of these antigen slides appear to be almost entirely
concordant with those established previously in RSIL by IFAT using
FYSA. Furthermore this antigen can be used with the same serum
dilutions and the same diagnostic criteria applied as have been used
for the FYSA.1
The other L. pneumophila serogroup 1 monovalent IFAT antigen (Zeus)

could also be used, but attention would have to be paid to the
interpretation of the endpoint and new (to the PHLS) diagnostic
criteria would need to be adopted.
Some laboratories may, for local organisational reasons, prefer to use

an EIA. Both EIAs examined performed adequately as screening
assays. However positive results would have to be confirmed using
one of the above L. pneumophila serogroup 1 IFAT kits: the combined
use of an EIA followed by a serogroup 1 IFAT is highly specific. None
of the L. pneumophila polyvalent antigens is sufficiently specific to
justify its routine use.
Cross-reactions due to campylobacter will continue to be a significant

problem whichever reagent is used. Consequently sera with
significantly positive titres should be referred to the reference
laboratory for further examination.
Background
Legionnaires’ disease (LD) is an uncommon form of pneumonia,

probably accounting for less than 5% of all pneumonia cases requiring
1 These slides will be available from The Binding Site from the end of August 1999.
It is likely that a complete ‘kit’ will be available during the following year. In the
mean time we would suggest that conjugate PF002, obtained from the same source,
be used in conjunction with the slides.

hospital admission (1). However the diagnosis and detection of cases of
LD is important for several reasons. Firstly, although uncommon
overall, legionella infections are much more significant in the context of
severe community-acquired pneumonias, accounting for 14-37% of
cases with an associated mortality rate in excess of 25% (2). Secondly,
the early recognition of cases of LD allows the source of actual, or
potential, outbreaks to be identified and hastens the implementation
of appropriate control measures.
As LD has no particular clinical features that clearly distinguish it from
other pneumonias, laboratory investigations must be relied upon if a
diagnosis is to be established with any confidence. A range of
approaches may be taken to establish the diagnosis (3).
Culture and isolation of legionellae from respiratory specimens

remains the definitive technique for the diagnosis of legionella
infection. A positive culture result allows further investigation of
sources and routes of infection and epidemiological typing of patient
and environmental isolates. Although culture is relatively slow (3-10
days), and requires a high degree of expertise, this approach should
be undertaken wherever and whenever possible. It is particularly
important in the investigation of nosocomial legionellosis where
infection caused by serogroups other than L. pneumophila serogroup 1
may be encountered.
Direct fluorescent antibody staining of respiratory specimens is a very

rapid way to establish a diagnosis. There are excellent reagents
commercially available. Unfortunately the sensitivity of this technique
is very poor. Again this method is particularly useful in the rapid
diagnosis of nosocomial legionellosis.
It is now well accepted that the use of enzyme immunoassays (EIAs)

for the detection of L. pneumophila antigen in urine, allows a diagnosis
of legionnaires’ disease to be established early in the course of
infection (4-6).
RSIL has, for about eight years, provided a reference service for the
detection of L. pneumophila serogroup 1 antigen in acute phase urine
samples. This service, which utilises an in-house EIA, has become
established as the most useful approach to the acute diagnosis of
legionella infections in the UK (6).
In 1997 almost two-thirds of the patients with a confirmed diagnosis

of L. pneumophila infection had a positive urinary antigen test. For
those patients for whom data are available, the mean time from
hospital admission to establishing a diagnosis of LD was 2.9 days
(range 0–23 days). In all cases the first collected specimen was
positive: in 22/100 cases this was on the day of admission (7).

Recently L. pneumophila urinary antigen assays have become available
from commercial sources. These assays have been shown to be highly
specific and sufficiently sensitive for routine diagnostic use (8,9).
Currently two EIAs and one immuno-chromatographic test (ICT) are
available2.
However, the estimation of serum antibody levels remains the most
widely used diagnostic method. This approach does not often allow a
diagnosis to be established during acute illness, and so has little value
in patient management or in the early detection of outbreaks, however
it is an important tool for case finding during outbreak investigations
and for late or retrospective diagnosis.
In England and Wales serological confirmation of suspected LD has
traditionally relied upon reagents supplied by the PHLS. Such supply
has been direct, by sale of serodiagnostic reagents for use in PHLS
and NHS laboratories and indirect by reference testing at CPHL.
However following the PHLS Board’s decision that from July 1994
reagent production at CPHL would cease, stocks of Legionella
pneumophila antigens have become depleted: L.pneumophila formalin-
killed yolk-sac antigen (FYSA) is no longer available and supply of
rapid microagglutination (RMAT) antigen will cease at the end of July
1999.
To allow all PHLS laboratories to make an informed choice when

selecting commercially available replacements, RSIL undertook an
evaluation of suitable commercial alternatives. This report presents
the results of this study and is intended to guide PHLS purchasing of
such reagents.
It should be noted that this study only sought to address the

laboratory diagnosis of L. pneumophila infection requiring admission
to hospital
2 The Binax EIA and ICT are available from Launch Diagnostics (01474 874426) and
the Biotest EIA from Biotest (0121-722 3393).

Legionella pneumophila assays evaluated
We attempted to identify all kits and reagents, intended for use in the
estimation of L. pneumophila antibody levels, that were commercially
available in the UK, either directly from the manufacturer or through
a UK supplier. Tables 1 and 2 show the kits and reagents that were
considered for inclusion in this study.
Previous work has shown that all subclasses of Ig must be determined

to maximise the diagnostic yield (3,10). Consequently assays designed
to determine specific Ig subclasses were not considered suitable for
evaluation. Similarly where a manufacturer produced a range of
assays designed to detect different legionella serogroups and/or
species only a selection of these were evaluated. Kits and reagents
selected for evaluation are shown in Table 1. Assays identified but not
evaluated are shown in Table 2.

Table 1. Legionella pneumophila assays evaluated
Assay Name Maufacturer/Source UK distributor Product

number number
1 Zeus Legionella ELISA Test System Zeus Scientific Inc. Quest Biomedical 4ZS20002
2 Sigma Diagnostic Legionella IgG/IgM/IgA Sigma Diagnostics Sigma Diagnostics EIA538-B
3 Zeus Legionella IFAT System (Gp 1) Zeus Scientific Inc. Quest Biomedical 4ZS20125
4 Zeus Legionella IFAT System (Gp 1-6) Zeus Scientific Inc. Quest Biomedical 4ZS20127
5 MRL Diagnostics Legionella IFAT (Heat-killed) MRL Diagnostics The Binding Site N/A
6 MRL Diagnostics Legionella IFAT (Formalin-killed) MRL Diagnostics The Binding Site N/A
7 Gull Laboratories L.pneumophila IFAT (Gp 1-6) Gull Laboratories GmbH Launch Diagnostics 4873131
Notes:
The Sigma Diagnostic Legionella IgG/IgM/IgA is originally sourced from Zeus Scientific but is presented differently (reagent buffer volumes etc.) and so
was evaluated separately.
Rather than use the currently available MRL Diagnostic Legionella IFAT slides we took the opportunity to evaluate the ‘trial’ slides being evaluated by MRL
as likely replacements to their current reagents. Following this evaluation the formalin-killed antigen slides will be made available from MRL from August
1999.

Table 2. Legionella pneumophila assays identified but not evaluated
Assay Name Maufacturer/Source UK distributor Product

number Number
8 Zeus Legionella IFAT System (Gp 1-4) Zeus Scientific Inc. Quest Biomedical 4ZS20126
9 Gull Laboratories L. pneumophila IFAT (Gp 1-8) Gull Laboratories GmbH Launch Diagnostics 4873171
10 Virion Serion L. pneumophila 1-7 IgG (quantitative) Serion GmbH The Binding Site ESR 106 G
11 Virion Serion L. pneumophila 1-7 IgM (quantitative) Serion GmbH The Binding Site ESR 106 M
12 Biolisa L. pneumophila 1-7 IgG Biomed Diagnostics GmbH None currently 108 223
13 Biolisa L. pneumophila 1-7 IgM Biomed Diagnostics GmbH None currently 108 224

Table 3. Assay information
Assay name Format Ig Class Specimen Antigens

volume
Zeus Legionella ELISA Test System EIA, 12x 1x8 well strips / Kit G/M/A 10ul Formalin-inactivated, sonicated preparation of L.
pneumophila Sgp1-6
Sigma Diagnostic Legionella EIA, 12x 1x8 well strips / Kit G/M/A 10ul Formalin-inactivated, sonicated preparation of L.
IgG/IgM/IgA pneumophila Sgp1-6
Zeus Legionella IFAT System IFAT, 10x 8 well slides / Kit G/M/A 10ul Heat-killed L. pneumophila Sgp 1
Zeus Legionella IFAT System IFAT, 10x 8 well slides / Kit G/M/A 10ul Heat-killed L. pneumophila Sgp1-6
MRL Diagnostics Legionella IFAT IFAT, 12 well slide (three G/M/A 10ul Heat-killed antigens of L.pneumophila Sgp 1, Sgp
Heat-killed (slides only) antigen spots per well) 3,4,5,6,8 and Legionella non-pneumophila (L.
bozemanii, micdadei, dumoffii, longbeachae Sgp 1&2)
MRL Diagnostics Legionella IFAT IFAT, 12 well slide (three G/M/A 10ul Formalin- killed antigens of L.pneumophila Sgp 1, Sgp
Formalin-killed (slides only) antigen spots per well) 3,4,5,6,8 and Legionella non-pneumophila (L.
bozemanii, micdadei, dumoffii, longbeachae Sgp 1&2)
Gull Laboratories L.pneumophila IFAT IFAT, 15x 10 well slides / Kit G/M/A 10ul Heat- inactivated L.pneumophila Sgp 1-6
(Gp 1-6)

Table 4. Manufacturer’s statement of claims, limitations and interpretation of each assay (from product insert where available)
Assay name Claims Negative Positive results Limitations and interpretation

results
Zeus Legionella an in-vitro diagnostic kit OD/Cutoff OD (OD OD/Cutoff OD a diagnosis should not be made on the basis of anti-legionella results alone. Test results for anti-legionella
ELISA Test for the qualitative Ratio <=0.90) (OD Ratio 0.91 -1.09) should be interpreted in conjunction with the clinical evaluation and the results of other diagnostic procedures
System detection of total
antibody (IgG/IgM/IgA) a positive result suggests infection with one or more of the group 1-6 species; however, one will not be able
to Legionella to distinguish between species with the results of this ELISA test alone.
AND pneumophila serogroup
the use of haemolytic, lipaemic, bacterially contaminated or heat inactivated specimens should be avoided.
1-6 in human sera.
Erroneous results may occur.
Sigma cross-reactivity may occur in sera with infections due to other legionella species.
Diagnostic
Legionella a negative result does not rule out the possibility of infection with legionella. Serum specimens taken too
IgG/IgM/IgA early during the course of infection may not yet have significant antibody titres. Some culture positive cases
of legionella do not develop antibody to legionella.
positive results may be due to cross reactivity with antibody generated as a result of non-legionella infection.
Serologic cross-reactivity has been reported with P.aeruginosa, several Rickettsia species, Coxiella burnetii
enteric gram-negative rods, Bacteroides species, Haemophilus species, Citrobacter freundii and
Campylobacter jejuni. Therefore, a positive result alone does not indicate infection with legionella.
Additionally, some reports indicate that a number of apparently healthy individuals may carry antibodies to
legionella; however a positive result, along with clinical signs and symptoms may indicate possible legionella
infection. Additional serologic testing, such as paired sera analysis by IFA, or other clinical testing such as
direct FA and culturing may be necessary to establish diagnosis.
the assay performance characteristics have not been established for matrices other than sera.
the affinity and\or avidity of the anti-IgG/IgM/IgA conjugate has not been determined.
although the conjugate is designed to detect human IgG, IgM and\or IgA, one will not be able to determine
which antibody is present with this assay.
early antibiotic therapy may suppress antibody response and some individuals may not develop antibodies
above detectable limits.
a single positive result only indicates previous immunologic exposure; level of antibody response may not be
used to determine active infection.
use of serogroups 1-6 for assessing antibody responses to different legionella species and serogroups has
not been established..Some infected patients may not have detectable levels of antibodies with this assay.
Four to eight weeks may be needed to detect an antibody response and antibody levels can fall to
undetectable levels within a month of infection.
specimens with OD ratio values in the equivocal range (0.91-1.09) should be retested. Specimens which
remain equivocal after repeated testing should be tested by an alternate serologic procedure such as the
Zeus Scientific Inc. Legionella IFA test system. An equivocal result does not exclude infection

results
Zeus Legionella an in vitro diagnostic <128 Manufacturer’s endpoint considerable experience in reading endpoints against the polyvalent antigen may be required to obtain the
Gp 1 IFAT assay for the detection is 1+ florescence - same titres as those obtained with monovalent antigens. Therefore polyvalent antigen titres should not be
System of L.pneumophila ‘barely visible staining’. used unless user proficiency can first be demonstrated.
antibodies in human
serum. A four-fold rise in titre a single test should not be used as the only criterion for diagnosis. The patient’s clinical data and other
AND >=128 from the acute to laboratory tests should be carefully reviewed by a medical authority before a diagnosis is made.
The specificity of this the convalescent phase
Zeus Legionella IFA test is enhanced provides evidence of a
Gp 1-6 IFAT when paired sera from recent infection.
System patients with symptoms
of legionellosis are A standing or single titre
tested. >=256 provides
presumptive evidence of
infection single titres of
less than 256 are not
considered evidence of
infection.
MRL <128 Not defined by the
Diagnostics manufacturer but the
Legionella IFAT endpoint was taken as
Heat-Killed 1+ fluorescence which is
(slides only) ‘definite yellow-green
staining’
A four-fold rise in titre

from <128 to >=256
provides evidence of a
recent infection.
A standing or single titre

>=256 provides
infection at an
undetermined time
Single titres of less than

128 are not considered
evidence of infection

results
MRL <16 Not defined by the
Diagnostics manufacturer but the
Legionella IFAT endpoint was taken as
Formalin-Killed 1+ fluorescence which is
(slides only) ‘definite yellow-green
staining’
A four-fold rise in titre

from <16 to >=64
provides evidence of a
recent infection.
A standing or single titre

>=128 provides
infection at an
undetermined time.
Single titres of less than

16 are not considered
evidence of infection
Gull detection of antibodies <128 Endpoint not defined. 4+ Cross-reactivity with different serogroups is not unusual. The pool antigen is therefore suitable for screening and
Laboratories to Legionella fluorescence is defined for exclusion of negative sera.
L.pneumophila pneumophila serogroups as about 50% of
IFAT (Gp 1-6) in serum samples bacteria exhibit strong
fluorescence.
A titre >=128 provides

evidence of a recent
legionella infection.
NOTES:
For the Zeus Legionella Gp 1 and Gp 1-6 IFAT kits in addition to the manufacturer’s defined endpoint a revised definition was also used. This was equivalent to ‘2+’ fluorescence as defined by the manufacturer.
The MRL slides were not supplied with product inserts as they were prepared for RSIL as ‘trial reagents’. The criteria adopted and presented here are those originally proposed by CDC Atlanta for the IFAT using
heat-killed antigen or those used by CPHL for the formalin-killed antigen.

Methods
An ‘outline’ evaluation protocol was circulated to the manufacturers/

distributors of each kit selected for evaluation (Appendix 1) and they were
invited to supply kits/reagents for the evaluation. Kits were supplied free
of charge or for a token charge in all cases.
A single batch of each kit/reagent was supplied and used throughout the
study3. Consequently no judgement can be made concerning the batch to
batch variation of the kits evaluated.
Each assay was tested against the whole panel of 932 sera. These sera
were collected from UK blood donors, proven cases of L. pneumophila
infection, proven cases of non-legionella pneumonia and a small number of
patients with undefined respiratory infection whose sera had been shown
to give falsely positive results in legionella assays.
Where provided, assays were performed and interpreted according to the

manufacturer’s product inserts. Any deviations from these instructions
are noted in the text and in the appropriate Technical Appraisal. For the
two MRL ‘trial’ slides the IFAT was undertaken using an in-house protocol
(Appendix 2).
The initial result given by each kit evaluated was used to assess the
specificity and sensitivity of each kit. The exceptions to this were the two
EIAs. Here if a major discrepancy was noted between the kits (positive vs.
negative), both assays were re-tested twice more and the consensus result
(2 of 3) was recorded. Also, as recommended by the manufacturer, sera
found to be ‘equivocal’ on initial testing were re-tested and the re-test
result was recorded. For estimates of specificity repeatedly ‘equivocal’
results were classified as ‘positive’.
Sera from the pneumonia patients had previously been examined by the
IFAT using FYSA. Although, due to antigen shortage, this assay was not
included in this evaluation the original results were used to obtain an
estimate of the concordance between the IFAT using FYSA and each of the
kits under evaluation (see Appendix 3). Clearly this analysis must be
treated with some caution as the titres of some sera might have declined
slightly since they were originally examined using FYSA.
3 However kits of a different batch were used for a small number of specimens to finish
the study in the cases of the Sigma EIA and the Zeus IFAT Gp 1 and Gp 1-6.

Specimen panels
Specimens for analysis were obtained from RSIL’s serum collection. Sera
had been stored at 4oC for not more than six months or at temperatures
between –70oC and –40oC for up to seven years before testing.
A. Control Groups
Blood donors
Donors were selected by age and sex, with an excess of males included to
take account of the ‘typical’ profile of legionnaires’ disease patients. Single
serum samples from 500 blood donors were examined as shown in Table
5.
Pneumonia patients with infections of known aetiology
Sera were available from 114 non-legionella pneumonia patients with

proven infection: 29 patients had three specimens available, 47 had paired
sera and 38 patients had a single serum available (Table 6.).
Patients with falsely positive legionella serology
Sera were available from 11 patients that gave positive results in the IFAT
(using FYSA) on initial testing, but were negative on re-testing in the
presence of ‘campylobacter blocking fluid’ (11,12). All 14 sera from these
patients (including three sera initially found to be negative) were examined
in this study.
B. Patients with proven L. pneumophila infection
Sera for the study were available from 117 patients with proven
L. pneumophila infection. All sera had previously been examined by IFAT
using FYSA antigen.
Proven L. pneumophila serogroup 1 infection
At least one serum specimen was available for examination in the study
from 111 patients with proven L. pneumophila serogroup 1 infection. The
diagnosis was established by culture of the organism in 66 cases and by
detection of L. pneumophila serogroup 1 urinary antigen in 45 cases. These
patients were categorised into four groups based on the serological
findings obtained on previous testing in the IFAT using FYSA (Table 7.).
Proven L. pneumophila non-serogroup 1 infection
Six patients with infection proven by culture of L. pneumophila of

serogroups other than serogroup 1 had serum specimens available for
testing (Table 8.).

Table 5. Age and sex distribution of the blood donor sera examined
Age range Male Female Total
30-39 100 66 166

40-49 100 66 166
50-59 100 68 168
Total 300 200 500
Table 6. Sera examined from patients with non-legionella pneumonia

of known aetiology
Causative Agent No. of Patients No. of Sera
Chlamydia spp. 16 27
C. burnettii 2 5
Flu A 2 4
Flu B 4 11
M. pneumoniae 37 68
RSV 2 6
S. pneumoniae (antigen in urine) 27 53
S. pneumoniae (blood culture) 23 44
S. pneumoniae and RSV 1 1
Totals 114 219

Table 7. Evidence of infection and serological findings (IFAT using
FYSA) for 111 patients with proven L. pneumophila serogroup 1
infection
Serological findings Evidence of infection Totals

(IFAT using FYSA) Culture Antigen No. of No. of
proven proven patients sera
Diagnostic rise in titre 20 17 37 89

Single/standing diagnostic titre 17 9 26 31
Positive but not diagnostic serology 17 13 30 40
Negative 12 6 18 28
Totals 66 45 111 188
Table 8. Infecting L. pneumophila serogroup and number of sera

available for six patients with proven L. pneumophila non-serogroup 1
infection
L. pneumophila serogroup No. of Patients No. of Sera
3 1 2
5 1 2
6 2 5
8 2 2
Total 6 11

Results
Analysis of the results is intended to allow the reader to draw their own
conclusions from the data. The results, which are presented in Tables 9 to
22, are summarised below.
Specificity in sera from apparently healthy people
Few, or no, data are available for commercial kits that specifically relate to
the situation in England and Wales. Consequently the first part of the
study was intended to determine the background seroprevelance estimated
by each of the various kits.
Sera from 500 blood donors (Table 5.) were examined and the results used
to estimate the specificity of each assay as applied to apparently health
populations. The titres obtained are shown in Table 9. and the specificity
of each assay, calculated from these data, is shown in Table 10.
It is immediately apparent that the assays show considerable differences

with estimated specificities ranging from 80.8%, for the Zeus Sgp 1-6 IFAT
antigen, to 98.8% for the MRL formalin Sgp 1 antigen. However when the
manufacturer’s definition for the endpoint was reviewed for the Zeus IFAT
antigens, it was clear that these were at variance with common practice
being defined as ‘1+ fluorescence, where 1+ = barely visible staining’. If the
manufacturer’s definition for ‘2+’ intensity (definite but dim staining) is
taken as the endpoint then the specificity was very much better being
97.4% for the polyvalent and 99.8% for the monovalent IFAT antigens
respectively.
Consequently for all subsequent studies both Zeus IFAT kits were read
according to the manufacturer’s definitions of ‘1+’ fluorescence, referred to
as ‘original’ reading and by their definition of ‘2+’ fluorescence, referred to
as ‘revised’ reading. For the sensitivity studies only the ‘revised’ ‘2+’
fluorescence endpoint results were recorded.
It had been intended only to further examine kits which had a specificity of
>95% for the blood donor population (Appendix 1). Taking into account
the revised reading of Zeus IFAT antigens all kits meet this criterion except
the Gull 1-6 IFAT antigen. However it was decided to include the Gull 1-6
assay for completeness.
Specificity and Sensitivity
The specificity and sensitivity of each kit was then determined in the
context of the differential diagnosis of pneumonia. Each of these
parameters was determined in two ways: firstly for ‘any positive result’
irrespective of whether or not the result would be considered diagnostic of
L. pneumophila infection. This was a specimen based analysis and was
intended to give an indication, for each assay, of the likely significance that
the testing laboratory should place on any positive result it might detect.

The second approach was to determine the specificity and sensitivity for
each kit with regard to a ‘diagnostic result’. This was a patient based
analysis that took into account all sera for a particular patient. For most
of the IFATs diagnostic results were based on either a single high titre or a
rise in titre between paired sera (as determined from the manufacturer’s
criteria detailed in Table 4). However for the two EIAs and the Gull 1-6
IFAT ‘any positive result’ and ‘diagnostic result’ were the same.
Specificity
The 219 sera from the 114 non-legionella pneumonia patients (Table 6.)
were all examined by each assay. The specificities for ‘any positive result’
ranged from 68.4% to 99.1% (Table 11) and are in good agreement with the
blood donor data. If the ‘original’ reading for the Zeus IFAT antigens are
disregarded the range is from 89.5% for Gull 1-6 to 99.1% for the Zeus 1
IFAT. Both EIAs gave specificities of 98.2%.
Again if the ‘original’ reading for the Zeus IFAT antigens are disregarded
the specificities for a ‘diagnostic result’ ranged from 89.5% for the Gull 1-6
to 100% for the MRL formalin IFAT antigens (Table 12).
Sensitivity - for L. pneumophila serogroup 1
Estimates of sensitivity were made using sera obtained from 111 patients
with proven L. pneumophila serogroup 1 infection (Table 7.). Of the 188
sera available 122 gave a positive result in at least one of the kits under
evaluation. The data for these 122 sera were therefore used to estimate
the sensitivity for ‘any positive result’ (irrespective of whether or not the
result would be considered diagnostic of infection) of each kit. The
estimates ranged from 69.7% for the MRL heat-killed IFAT antigen to
95.1% for the Gull 1-6 IFAT antigen (Table 13). The high sensitivity of the
Gull 1-6 antigen is not unexpected given its correspondingly low
specificity.
Of the 111 patients 92 were found to have serology results diagnostic of LD

by at least one of the kits under evaluation. Results for these patients were
used to estimate the sensitivity for a ‘diagnostic result’ for each kit (Table
14). The estimates ranged from 63.0% for the MRL IFAT antigens to 97.8%
for the Gull 1-6 IFAT.
As the two EIAs are essentially the same assay the different estimates of
sensitivity for each of them was unexpected. Some of the sera which were
negative in the Sigma EIA gave repeatedly ‘equivocal’ results in the Zeus
EIA and hence were considered as ‘positive’ for the purpose of this
evaluation. Most of these sera had OD ratios close to the cut-off value of
3.0 in the Sigma EIA (see the Technical Appraisal for further comment).
Sensitivity - for L. pneumophila non- serogroup 1

With so few sera available from patients with culture proven
L. pneumophila infection caused by serogroups other than serogroup 1
(Table 8), it is difficult to draw any conclusions. However it can be seen
that the antibody response it not always serogroup specific as the
monovalent serogroup 1 antigens detected antibodies in at least some of
the non-serogroup 1 patients. Conversely it can be seen that the
polyvalent IFAT antigens detected antibodies that were not detected by the
serogroup 1 antigens (Table 15).
Estimates of the Positive and Negative Predictive Values
The Positive Predictive Value (PPV) is the chance that a positive result is a
true, rather than a false, positive and the Negative Predictive Value (NPV)
is the chance that a negative result is truly negative and not a false
negative. Legionella infection is estimated to be responsible for between
2% and 5% of community-acquired pneumonia requiring admission to
hospital in the UK (1,2), and for such low prevalence infections the PPV is
very dependant on the specificity of the assay being used while in contrast
the NPV is very high almost irrespective of test specificity.
PPV and NPVs were calculated for ‘any positive’ result (Table 16.) and for a
‘diagnostic’ result (Table 17.) for both 2% and 5% prevalence. It can be
seen that for the EIAs the PPV of a diagnostic result is somewhere between
48% and 73.2%. For the Sgp1 heat-killed antigens (MRL and Zeus) it is
between 58.8% and 80.5% and for the MRL formalin Sgp 1 antigen it is
estimated to be 100%. However for the polyvalent IFAT kits it ranges from
as low as 16.0% for the Gull 1-6 to 60.3% for the Zeus 1-6.
Campylobacter cross-reactions
Falsely positive results in legionella serology due to camplyobacter

infection are now recognised as a significant, albeit infrequent, problem
(11,12). Table 18. shows that all the kits gave positive results with sera from
some of the patients and all gave diagnostic titres with sera from at least
one of the patients. This confirms previous work that the cross-reaction is
not dependent upon the type of antigen or assay format used.
Sequential testing
It is likely that, at least for some laboratories, it might be logistically

advantageous if an EIA can be used for L. pneumophila antibody
estimation. However it is clear from the data presented in Tables 16. and
17. that the PPVs of the EIAs are not sufficiently high for either EIA to be
depended upon as the sole diagnostic method.
Consequently we looked at the possibility of using an EIA as a screening

test followed by one of the IFATs as a confirmatory test for any samples
found to be EIA positive. The sensitivity, specificity, PPV and NPV was
calculated for such sequential testing and the results are shown in Table

19. for ‘any positive result’ and in Table 20. for a diagnostic result. Clearly
this approach gives very high PPVs but the sensitivity of sequential testing
was a little lower than single assay testing.
Effect of changing the cut-off value on the performance of the

Gull 1- 6 IFAT.
As far as possible the kits under evaluation were used as indicated by the
manufacturers. However from the outset it was clear that the Zeus IFAT
antigens did not perform well if the manufacturer’s definition for the
endpoint was used. For the Zeus reagents this was overcome by redefining
the endpoint (see results for seroprevalence above) and the redefined
assays performed quite well. In contrast the Gull 1-6 IFAT gave good
fluorescence and the endpoint was clear if the manufacturer’s instructions
were followed but the specificity of the assay was very poor.
It can be seen from Table 9. that 35 of the 46 blood donor sera that were
found to be positive were only positive at a titre of 128. Consequently the
performance of the Gull 1-6 IFAT was re-evaluated with different titres
being used as the ‘cut-off’. If a cut-off of 256 is adopted the specificity of
the assay improves considerably (Tables 21.) and this is reflected in
improved PPVs (Table 22.) which are slightly better than those obtained
with the Zeus 1-6 IFAT. Increasing the cut-off titre to 512 improves the
specificity only slightly further but the sensitivity falls sharply.

Table 9. Titres obtained for each IFAT assay using sera from 500 blood donors
Titre Zeus 1 Zeus 1 MRL 1 Zeus 1-6 Zeus 1-6 MRL 3-8 Gull 1-6 Titre MRL 1 MRL 3-8
(original) (revised) Heat (original) (revised) Heat Formalin Formalin
<1:128 458 499 476 404 487 476 454 <1:16 494 482
1:128 30 1 17 39 11 17 35 1:16 4 7
1:256 12 0 4 57 2 4 8 1:32 1 5
1:512 0 0 3 0 0 3 3 1:64 1 3
1:1024 0 0 0 0 0 0 0 1:128 0 3

Table 10. Specificity of all assays determined using sera from 500 blood donors
EIA4 IFAT
Sigma 1-6 Zeus 1-6 Zeus 1 MRL 1 Zeus 1-6 MRL 3-8 Gull 1- 6
(original) (revised) Heat Formalin (original) (revised) Heat Formalin
Cut off titre 128 128 128 16 128 128 128 16 128
No. Positive 10 13 42 1 24 6 96 13 24 18 46
No. Negative 490 487 458 499 476 494 404 487 476 482 454
Specificity 98.0 97.4 91.6 99.8 95.2 98.8 80.8 97.4 95.2 96.4 90.8
95% CI - lower limit 96.4 95.7 88.8 98.9 92.94 97.4 77.1 95.7 92.9 94.4 87.9
95% CI - upper limit 99.0 98.6 95.9 100.0 96.9 99.6 84.2 98.6 96.9 97.9 93.2
4Of the 10 Sigma EIA positive sera one was repeatedly ‘equivocal’. Of the 13 Zeus EIA positive sera four were repeatedly ‘equivocal’. A further
10 (Sigma) and 11 (Zeus) sera were ‘equivocal’ on initial testing but were ‘negative’ on re-testing.

Table 11. The number of non-legionella pneumonia patients whose sera gave a positive result in the assays.
Agent (no. of pateints) EIA5 IFAT

Chlamydia spp. (16) 0 0 2 0 3 1 10 0 3 3 2

C. burnettii (2) 0 0 0 0 0 0 0 0 0 0 0
Flu A (2) 0 0 0 0 0 0 0 0 0 0 0
Flu B (4) 0 0 0 0 0 0 0 0 0 0 0
M. pneumoniae (37) 0 0 5 1 2 1 12 3 1 1 6
RSV (2) 0 0 0 0 0 0 1 0 0 0 0
S. pneumoniae (antigen in urine) (27) 1 1 0 0 0 0 7 0 0 0 0
S.pneumoniae (blood culture) (23) 1 1 1 0 2 0 6 1 3 1 4
S.pneumoniae and RSV (1) 0 0 0 0 0 0 0 0 0 0 0
Total no. positive (114) 2 2 8 1 7 2 36 4 7 5 12
Specificity (%) 98.2 98.2 93.0 99.1 93.9 98.2 68.4 96.5 93.9 95.6 89.5
95% CI - lower limit 93.8 93.8 86.6 95.3 87.8 93.8 59.1 91.3 87.8 90.1 82.3
95% CI - upper limit 99.8 99.8 96.9 100.0 97.5 99.8 76.8 99.0 97.5 98.6 94.4
5 The sera from the ‘S.pneumoniae antigen positive’ patient were repeatedly ‘equivocal’.

Table 12. The number of non-legionella pneumonia patients whose sera gave diagnostic results in the assays.
Agent (no. of patients) EIA IFAT

Chlamydia spp. (16) 0 0 1 0 0 0 1 0 0 0 2

C. burnettii (2) 0 0 0 0 0 0 0 0 0 0 0
Flu A (2) 0 0 0 0 0 0 0 0 0 0 0
Flu B (4) 0 0 0 0 0 0 0 0 0 0 0
M. pneumoniae (37) 0 0 1 1 1 0 5 2 0 0 6
RSV (2) 0 0 0 0 0 0 1 0 0 0 0
S. pneumoniae (antigen in urine) (27) 1 1 0 0 0 0 0 0 0 0 0
S. pneumoniae (blood culture) (23) 1 1 0 0 0 0 1 1 0 1 4
S. pneumoniae and RSV (1) 0 0 0 0 0 0 0 0 0 0 0
Total no. positive (114) 2 2 2 1 1 0 8 3 0 1 12
Specificity (%) 98.2 98.2 98.2 99.1 99.1 100 93.0 97.4 100.0 99.1 89.5
95% CI - lower limit 93.8 93.8 93.8 95.3 95.3 96.8 86.6 92.5 96.8 95.3 82.3
95% CI - upper limit 99.8 99.8 99.8 100.0 100.0 100.0 96.9 99.5 100.0 100.0 94.4

Table 13. – Sensitivity of ‘any positive result’ for each kit for patients with proven L. pneumophila serogroup 1
infection
EIA IFAT
Zeus 1-6 Sigma 1-6 Zeus 1 MRL 1 Zeus 1-6 Gull 1-6
(revised) Heat Formalin (revised)
No. of sera positive 107 95 103 85 111 104 116

No. of sera negative 15 27 19 37 11 18 6
Relative Sensitivity (%) 87.7 77.9 84.4 69.7 91.0 85.2 95.1
95% CI lower 80.5 69.5 76.8 60.7 84.4 77.7 89.6
95% CI upper 93.0 84.9 90.4 77.7 95.4 91.0 98.2
NOTE: Of the 188 sera from patients with L.pneumophila serogroup 1 proven infection, 122 were found to be positive (any titre) by at least one of the kits under
evaluation. These sera were used to estimate the comparative sensitivity for any positive result for each kits and the results are shown in this table. Results for the
two MRL 3-8 antigens are not shown.

Table 14. – Sensitivity of a ‘diagnostic result’ for each kit for patients with proven L. pneumophila serogroup 1
infection
EIA IFAT
Number with diagnostic serology (92) 86 75 65 58 58 69 90
Relative sensitivity 93.5 81.5 70.7 63.0 63.0 75.0 97.8

95% CI lower 86.3 72.1 60.2 52.3 52.3 64.9 92.4
95% CI upper 97.6 88.9 79.7 72.9 72.9 83.4 99.7
NOTE: Of the 111 proven cases of L. pneumophila serogroup 1 infection, 92 were found to have diagnostic serology by at least one of the kits under evaluation. The table
shows the comparative sensitivity estimated for each assay for this group of patients. Results for the two MRL 3-8 antigens are not shown.

Table 15. Sensitivity of culture proven L. pneumophila non-serogroup 1 infection.
EIA IFAT
L.pneumophila
Sgp 3 (1) 0 0 0 0 0 1 1
Sgp 5 (1) 1 1 1 1 1 1 1
Sgp 6 (2) 0 0 0 0 0 0 0
Sgp 8 (2) 0 0 1 0 0 1 1
Total (6) 1 1 2 1 1 3 3

Table 16. Positive predictive values (PPV) and negative predictive values (NPV) of any positive result assuming
a prevalence of L. pneumophila serogroup 1 infection of 2% or 5%
PPV NPV
Assay name Sensitivity Specificity 2% 5% 2% 5%
(any positive result) (any positive result) (prevalence) (prevalence) (prevalence) (prevalence)
EIAs
Sigma 1-6 77.9 98.2 46.9 69.5 99.6 98.9
Zeus 1-6 87.7 98.2 49.9 71.9 99.7 99.4
IFAT Sgp 1
Zeus 84.4 99.1 65.7 83.2 99.7 99.2
MRL (Heat) 69.7 93.9 18.9 37.6 99.4 98.4
MRL (Formalin) 91.0 98.2 50.8 72.7 99.8 99.5
IFAT Sgp 1-6
Zeus 85.2 96.5 33.2 56.2 99.7 99.2
Gull 95.1 89.5 15.6 32.3 99.9 99.7
NOTE: Legionella infection is estimated to be responsible for between 2% and 5% of community-acquired pneumonia requiring admission to hospital in the UK. Positive and Negative predictive
values were therefore calculated for both these prevalence rates.

Table 17. Positive predictive values (PPV) and negative predicative value (NPV) of diagnostic results assuming
a prevalence of L. pneumophila serogroup 1 infection of 2% or 5%
PPV NPV
Assay name Sensitivity Specificity 2% 5% 2% 5%
(diagnostic result) (diagnostic result) (prevalence) (prevalence) (prevalence) (prevalence)
EIAs
Sigma 1-6 81.5 98.2 48.0 70.4 99.6 99.0
Zeus 1-6 93.5 98.2 51.5 73.2 99.9 99.7
IFAT Sgp 1
Zeus 70.7 99.1 61.6 80.5 99.4 98.5
MRL (Heat) 63.0 99.1 58.8 78.7 99.3 98.1
MRL (Formalin) 63.0 100.0 100.0 100.0 99.3 98.1
IFAT Sgp 1-6
Zeus 75.0 97.4 37.1 60.3 99.5 98.7
Gull 97.8 89.5 16.0 32.9 100.0 99.9
NOTE: NOTE: Legionella infection is estimated to be responsible for between 2% and 5% of community-acquired pneumonia requiring admission to hospital in the UK. Positive and Negative
predictive values were therefore calculated for both these prevalence rates.

Table 18. Results obtained using 14 sera, from 11 patients, that gave positive results in the IFAT (using FYSA)
on initial testing, but were negative on re-testing in the presence of ‘campylobacter blocking fluid’.
EIA IFAT
Diagnostic serology (2) 5 5 3 4 1 2 9
Positive but not diagnostic (9) 2 2 0 2 9 1 N/A
Negative (0) 4 4 8 5 1 8 2
Totals number with any positive 7 7 3 6 10 3 9

serology

Table 19. Sequential testing - EIA followed by a second confirmatory test for samples yielding a positive EIA
result. Positive predictive values (PPV) and negative predicative value (NPV) of ‘any positive result’ assuming a
prevalence of L. pneumophila serogroup 1 infection of 2% or 5%
PPV NPV
Confirmatory assay Sensitivity Specificity 2% 5% 2% 5%
(positive result) (positive result) (prevalence) (prevalence) (prevalence) (prevalence)
IFAT Sgp 1
Zeus 82.0 100.0 100.0 100.0 99.6 99.1
MRL (Heat) 67.2 100.0 100.0 100.0 99.3 98.3
MRL (Formalin) 89.3 100.0 100.0 100.0 99.8 99.4
IFAT Sgp 1-6
Zeus 79.5 99.1 64.3 82.3 99.6 98.9
Gull 85.4 99.1 65.9 83.3 99.7 99.2
values were therefore calculated for both these prevalence rates. Any samples repeatedly ‘equivocal’ in the EIA were considered to be ‘positive’ and were therefore examined in the confirmatory
test.

Table 20. Sequential testing - EIA followed by a second confirmatory test for any sample yielding a positive
EIA result. Positive predictive values (PPV) and negative predicative value (NPV) of a ‘diagnostic result’ assuming a
prevalence of L. pneumophila serogroup 1 infection of 2% or 5%
PPV NPV
Confirmatory assay Sensitivity Specificity 2% 5% 2% 5%
(diagnostic result) (diagnostic result) (prevalence) (prevalence) (prevalence) (prevalence)
IFAT Sgp 1
Zeus 62.0 100.0 100.0 100.0 99.2 99.8
MRL (Heat) 57.0 100.0 100.0 100.0 99.1 97.8
MRL (Formalin) 56.0 100.0 100.0 100.0 99.1 97.7
IFAT Sgp 1-6
Zeus 67.0 99.1 60.3 79.7 99.3 98.3
Gull 84.0 99.1 65.6 83.1 99.7 99.2
values were therefore calculated for both these prevalence rates. Any samples repeatedly ‘equivocal’ in the EIA were considered to be ‘positive’ and were therefore examined in the confirmatory
test.

Table 21. Effect of changing the ‘cut-off’ titre of the Gull 1-6 IFAT kit on the specificity obtained with blood
donor sera
Cut off titre

128 256 512
No. of sera positive 46 11 3

No. of sera negative 454 489 497
Specificity 90.8 97.8 99.4

95% CI - lower limit 87.9 96.1 98.3
95% CI - upper limit 93.2 98.9 99.9

Table 22. Positive (PPV) and Negative (NPV) predicative value of diagnostic results assuming a prevalence of L.
pneumophila serogroup 1 infection of 2% or 5% - effect of changing the ‘cut-off’ titre for the Gull 1-6 IFAT kit.
PPV NPV
Gull 1-6 Sensitivity Specificity 2% 5% 2% 5%
Cut off titre (diagnostic result) (diagnostic result) (prevalence) (prevalence) (prevalence) (prevalence)
128 97.8 89.5 16.0 32.9 100.0 99.9

256 84.8 97.4 40.0 63.2 99.7 99.2
512 66.3 98.2 42.9 66.0 99.3 98.3
values were therefore calculated for both these prevalence rates.

Conclusions
The simplest option for the replacement of the CPHL L.pneumophila serogroup 1
FYSA is to substitute for it the MRL formalin-killed IFAT antigen. Results
obtained for the L. pneumophila serogroup 1 component of these antigen slides
appear to be almost entirely concordant with those established previously in
RSIL by IFAT using FYSA (see Appendix 3). Furthermore this antigen can be
used with the same serum dilutions and the same diagnostic criteria applied as
have been used for the FYSA.
The other L. pneumophila serogroup 1 monovalent IFAT antigen (Zeus) could

also be used, but attention would have to be paid to the interpretation of the
endpoint and new (to the PHLS) diagnostic criteria would need to be adopted.
Some laboratories may, for local organisational reasons, prefer to use an EIA.
Both EIAs examined performed adequately as screening assays. However
positive results would have to be confirmed using one of the above L.
pneumophila serogroup 1 IFAT kits: the combined use of an EIA followed by a
serogroup 1 IFAT is highly specific. None of the L. pneumophila polyvalent
antigens is sufficiently specific to justify its routine use.
Cross-reactions due to campylobacter will continue to be a significant problem

whichever reagent is used. Consequently sera with significantly positive titres
should be referred to the reference laboratory for further examination.
Acknowledgements
We would like to acknowledge and thank the following for contributing to this
evaluation. The North London Blood Transfusion Centre for providing the blood
donor sera; staff of the Atypical Pneumonia Unit, RSIL for support and
assistance; CPHL Central Services for providing staff support for the study;
PHLS Statistics Unit for advice, and the manufacturers and distributors of the
kits for provision of the kits.

References
1. British Thoracic Society. Community-acquired pneumonia in adults in British
hospitals in 1982-1983: A BTS/PHLS survey of aetiology, mortality, prognostic factors
and outcome. Quart J Med 1987;62:195-220.
2. Hubbard RB, Mathur RM, Macfarlane JT. Severe community-acquired legionella

pneumonia: treatment, complications and outcome. Quart J Med 1993;86:327-32.
3. Harrison TG, Taylor AG. eds. A laboratory manual for legionella. Chichester: John
Wiley and Sons; 1988.
4. Birtles R, Harrison TG, Samuel D, Taylor AG. Evaluation of urinary antigen ELISA
for diagnosing Legionella pneumophila serogroup 1 infection. J Clin Pathol
1990;43:685-90.
5. Kohler RB, Zimmerman SE, Wilson E, et al. Rapid radioimmunoassay diagnosis of

Legionnaires' Disease. Ann Intern Med 1981;94:601-5.
6. Fehrenbach FJ, Horbach I, Ruf B, Schürmann D, Pohle HD. Rapid detection of

Legionella antigen in tissues and body fluids. Israel J Med Sci 1986;22:706-10.
7. Joseph CA, Harrison TG, Ilijic-Car D, Bartlett CLR. Legionnaires’ disease in residents
of England and Wales: 1997. Commun Dis Public Health 1998;1:252-8.
8. Harrison T, Uldum S, Alexiou-Daniel S, Bangsborg J, Bernander S, Drasar V, Etienne

J, Helbig J, Linday D, Lochman I, Marques T, de Ory F, Tartakovskii I, Wewalka G,
Fehrenbach F. A multicenter evaluation of the Biotest legionella urinary antigen EIA.
Clin Microbiol Infect 1998;4:359-64.
9. Hackman BA, Plouffe JF, Benson RF, Fields BS, Breiman RF. Comparison of Binax
Legionella Urinary Antigen EIA kit with Binax RIA urinary antigen kit for detection of
Legionella pneumophila serogroup 1 antigen. J Clin Microbiol 1996;34:1579-80.
10. Wilkinson HW, Farshy CE, Fikes BJ, Cruce DD, Yealy LP. Measurement of
immunoglobulin G-, M- and A- specific titres against Legionella pneumophila and
inhibition of titres against nonspecific Gram-negative bacterial antigens in the indirect
immunofluorecence test for legionellosis. J Clin Microbiol 1979;10:685-9.
11. Boswell TC. Serological cross reaction between legionella and campylobacter in the
rapid microagglutination test. J Clin Pathol 1996;49:584-6.
12. Boswell TC, Marshall LE, Kudesia G. False-positive legionella titres in routine
clinical serology testing detected by absorption with campylobacter: implications for
the serological diagnosis of legionnaires' disease. J Infect 1996;2:23-6.

Technical Appraisal
Zeus Legionella ELISA Test System
General
The Zeus Legionella ELISA Test System is an antibody capture microplate EIA
for the detection of IgG, IgM and IgA antibodies to Legionella pneumophila.
Serum specimens are added to the formalin inactivated, solublised L.
pneumophila Sgp 1-6 antigens coated wells. Following incubation, peroxidase
conjugated goat anti-human IgG/A/M is added, which reacts with antibody
immobilized on the solid phase in the above step. This is then incubated with
peroxidase substrate solution that by hydrolysis reaction produces a colour
change.
The IgGMA test is supplied as a kit of 12x 1x8 wells strips, peroxidase
conjugated anti-human IgG/A/M, positive and negative controls, sample
diluent, substrate solution, stop solution and wash buffer concentrate. Each 8-
well strip is coated with a preparation of the antigen. Each time the assay is
run, the low positive standard must be run in triplicate. A high positive and
negative control and reagent blank must be included in each assay. A total of
six wells are required to run the controls which would result in considerable
extra cost when testing small batches. Washing stages are critical especially for
specimens that fall within the ‘equivocal’ range.
Ease of use
The EIA was simple to perform and the endpoint is objectively determined.
Total assay time was about 50 minutes most of this is ‘hands on’ time.
Sigma Diagnostic Legionella IgG/IgM/IgA
General
The Sigma Diagnostic Legionella IgG/IgM/IgA is manufactured by Zeus

Diagnostics for Sigma. The assay format, reagents and product inserts are
essentially the same.
Ease of use
In the format examined here the Sigma kit was supplied with larger volumes of
most reagents. In this evaluation the OD values obtained with the Sigma kit
appeared to be consistently higher than those obtained with the Zeus kit. It is
unclear whether or not this is due to batch to batch variation or not.
Perversely, because the cut-off value is calculated from the OD of the low
positive standard (LPS) the effect of this was that the overall sensitivity was

lower for the Sigma compared to the Zeus. However fewer samples gave
equivocal results which needed to be repeated.
Zeus Legionella IFAT System (Gp 1)
General
The Zeus Legionella IFAT System (Gp 1) is an antibody capture

immunofluorescence procedure for the detection of antibodies to Legionella
pneumophila serogroup 1. Serum specimens are added to the L. pneumophila
serogroup 1 antigen coated wells of the slide. Following incubation, polyvalent
anti-human FITC labelled globulin is added, which reacts with antibody
immobilized on the solid phase in the above step. This is then examined under
a fluorescence microscope for ‘apple-green’ staining.
The IFAT test is supplied as a kit of 10x 8-well slides, anti-human FITC-labelled
globulin, positive control, buffer concentrate and mounting fluid. Each 8-well
slide contains fixed L. pneumophila substrate (antigen). The positive and
negative controls must be used each time the assay is run.
Reading stages are critical. As noted in the product insert considerable

experience in reading may be required before consistent results can be
obtained. The endpoint was defined as ‘1+ fluorescence = barely visible
staining’. This is at variance with standard practice where endpoints are
typically defined as ‘definite but dim yellow-green staining’. Using the
manufacturer’s endpoint the assay performed badly however with revised
reading of the endpoint performance was good.
Although the test was simple to use, the instructions provided were not clear.
The quantity of serum dilution to be added to the well was not stipulated
(although we were subsequently provided with an amended method sheet).
Similarly the instructions stipulated ‘one drop’ of conjugate to be added, but
did not indicate the volume of the drop.
Components of the Kit need to be separated on arrival, as slides are stored at -

20°C but reagents at 2-8°C
Additional information
The L. pneumophila serogroup 1 antigen and all other kit components

(conjugate, positive control serum etc.) can be purchased separately. The
antigen is available as ‘wet’ antigen or as fixed slides.
Zeus Legionella IFAT System (Gp 1-6)

General
The Zeus Legionella IFAT System (Gp 1-6) is an antibody capture

pneumophila serogroups1-6. Serum specimens are added to the L. pneumophila
serogroup 1-6 antigen coated wells of the slide. Following incubation, polyvalent
The IFAT test is supplied as a kit of 10x 8-well slides, anti-human FITC-labelled
globulin, positive controls, buffer concentrate and mounting fluid. Each 8-well
slide contains fixed L. pneumophila substrates (antigens). Each time the assay
is run, the positive (total of 6) and negative controls are required to be run.
Reading stages important and as mentioned in the product insert considerable

experience in reading might be required before endpoints can be determined
with confidence. This problem is more extreme than seen with the monovalent
antigens as the number of bacteria fluorescing in pooled antigens varies
dependant upon which components of the pool are reacting.
The methods sheet provided is the same as for the monovalent antigen – this
can be confusing. All other comments are as for the Zeus Legionella IFAT
System (Gp 1) above.
MRL Diagnostics Legionella IFAT Heat-Killed (slides only)
General
The MRL Diagnostics Legionella IFAT is an antibody capture

pneumophila serogroups1, 3-8 and non-pneumophila. Serum specimens are
added to the antigen-coated wells of the slide. Following incubation, polyvalent
The IFAT test was supplied as a ‘trial’ batch of 12-well slides only. Each slide
well consisted of pools of three fixed Legionella substrates (antigens). In-house
positive control and a commercial conjugate (PF002 The Binding Site) were
used for this study. Each time the assay was run, the positive control was
included. Because of the orientation of the antigen pools within the well (in a
circle), a diagrammatic representation of the slide, as viewed under the
microscope, is essential.

Although intensity of the fluorescence was generally good, serum titrations
tended to ‘trail off’ which lead to differences in the interpretation of endpoints
depending on observer.
Most serum samples gave positive results with the Legionella non-pneumophila
pool antigen and these results were not considered further in this study. Data
obtained using this Legionella spp. antigen pool would have to be treated with
considerable caution.
MRL Diagnostics Legionella IFAT Formalin-Killed (slides only)
This is the same as the MRL heat-killed antigen discussed above with the
exception that the antigens are formalin killed.
Endpoints were easier to read, the outline of the organisms being clearly
delineated, and serum dilutions showed less ‘trailing off’ and consequently
more consistent determination of the endpoint.
Gull Laboratories L.pneumophila IFAT (Gp 1-6)
General
The Gull Laboratories L. pneumophila IFAT System is an antibody capture

pneumophila serogroup 1-6. Serum specimens are added to the antigen-coated
wells of the slide. Following incubation, polyvalent anti-human FITC labelled
globulin is added, which reacts with antibody immobilized on the solid phase in
the above step. This is then examined under a fluorescence microscope for
‘apple-green’ staining.
The IFAT test is supplied as a kit of 15x 8-well slides, negative and positive
controls, conjugate , mounting fluid and buffer concentrate. Each 8-well slide
contains fixed L. pneumophila substrate (antigen). Each time the assay is run,
the positive and negative controls have to be run. Although the test and
instructions were simple to use, the definition of an endpoint is not given in the
product insert. The interpretation of a positive result is unusual, as it appears
that a single positive titre of 128 is considered evidence of recent infection.

Appendix 1. - Outline protocol sent to UK distributors
RSIL – Evaluation of commercial legionella antibody estimation

reagents/kits
PHLS production and distribution of reagents for the estimation of antibodies

against Legionella pneumophila will be terminated shortly. To allow all PHLS
laboratories to make an informed choice when selecting commercially available
replacements, RSIL is undertaking an evaluation of suitable alternatives. The
results of the study will be used to guide PHLS purchasing of such reagents.
Scope of study
We propose to look at all kits/reagents commercially available in UK that might

be suitable alternatives to the current PHLS reagents. Kits intended specifically
to determine IgM or IgG have been excluded from the study as they are of little
diagnostic value.
Where manufacturers produce a selection of kits each of which detect a

different range of L.pneumophila serogroups or species, only a L.pneumophila
serogroup 1 kit and one other have been included.
Kits suggested for inclusion
IFAT: L. pneumophila Sgp1 FYSA (in-house kit for comparison),

Zeus L. pneumophila Sgp1,
Zeus L. pneumophila Sgps1-6,
MRL L. pneumophila/Legionella spp., (antigen slides only)
Gull Laboratories L. pneumophila 1-6
EIA: Sigma Legionella IgG/A/M

Zeus L. pneumophila Sgp 1-6
Outline of the study
Few, or no, data are available for these commercial kits which specifically relate
to the situation in England and Wales. Consequently the first part of the study
will seek to determine the background seroprevelance estimated by each of the
various kits.
Approximately 500 blood donor sera (age stratified) will be examined. For
indirect immunofluorescent antibody tests these sera will be examined at the
screening dilution recommended by the manufacturers’ and at half this dilution
(i.e. if the screening dilution is 1/128 we will examine sera at both 1/64 and
1/128 – positives will be titrated to endpoint).

Assuming the assays will exclude >95% of the normal (blood donor) population
at the manufacturer’s screening dilution we will go on to determine the
specificity and sensitivity of each kit in the context of the differential diagnosis
of pneumonia.
To determine sensitivity we will examine sera (paired where available) from L.

pneumophila serogroup 1 culture proven cases (>50), L. pneumophila urinary
antigen positive patients (>50) and from L. pneumophila serologically (LMR
IFAT) proven cases (>100).
Sera from 150 – 250 patients with microbiologically confirmed non-L.

pneumophla respiratory infections (M. pneumoniae, S. pneumoniae, Chlamydia
spp. etc.) will be examined to determine test specificity.
Data obtained
Data obtained will remain the property of the PHLS. It is our intention to make
the data, or a summary of it, available to interested parties. In the longer term
we will seek to publish this work in one of the peer reviewed microbiology
journals.

Appendix 2. The in-house protocol used for the MRL slides
Indirect Immunofluorescence Antibody Test (IFAT) for the detection of

antibodies to Legionella using the MRL IFAT heat and formalin-killed
antigen slides.
MATERIALS
1. MRL L. pneumophila slides 12 well heat-killed antigen slides

(3antigens/well).
2. MRL L. pneumophila slides 12 well formalin-killed antigen slides
(3antigens/well).
3. L. pneumophila Sgp1 positive control serum (human), (LMR - Code
BCS02885).
4. FITC labelled anti-human (IgG, IgM, IgA) immunoglobulin, (PF002 - The
Binding Site).
5. Phosphate Buffered Saline (PBS), (Unipath Ltd - BR14a) – to be prepared
before start of procedure.
TEST PROCEDURE
1. Remove slides from refrigerator and allow then to reach room temperature.
Label each slide clearly.
2. Doubling serum dilutions are prepared in PBS starting at 1:16 (for formalin-
killed antigens) and 1:64 (for heat-killed antigens). Serum samples are
examined at screening dilutions of either 1:16 and 1:32 (for formalin
antigens) or 1:64 and 1:128 (for heat-killed antigens) Any samples found to
be positive are then to titrated to their endpoint
3. Add 20ul of the serum sample and positive control at the appropriate
dilution to each well.
4. Incubate the slides in a wet box at 37°C for 30 minutes.
5. Rinse the slides with PBS and wash for 20 minutes in PBS. Rinse briefly
with distilled water. Gently blot dry using tissue and allow the slides to dry
at 37°C for 5-10 mins.
6. Add 10ul of the FITC conjugate at the predetermined working dilution (1/60)
to each well of the slide.
7. Incubate the slides in a wet box at 37°C for 30 mins.
8. Wash as outlined above.
9. Add mounting fluid (MRL - Bartonella kit) to the slide and place cover slip.
10. Examine the wells for fluorescence using an appropriate microscope
equipped with x40 water immersion lens and x15 eyepieces.
EXAMINATION OF SLIDES

1. Refer to the manufacturers diagram for the orientation of the antigens
within each well.
2. The fluorescence is scored ++, +, +/-, - . Fluorescence scored as ‘+’ is defined
as the end-point and as being of sufficient intensity for the organisms to be
seen clearly.
3. The in-house positive control human reference serum must titrated for each
run.
INTERPRETATION
The following criteria were adopted for the study:
For the MRL formalin-killed antigen slides;
1. A four-fold rise in titre to >=64, or a single/standing titre of >=128 was

taken as diagnostic evidence of a L. pneumophila infection.
2. A single titre of >=16<128 was considered positive but not diagnostic of
L. pneumophila infection.
3. Single titres of less than 16 are considered negative.
For the MRL heat-killed antigen slides;
1. A four-fold rise in titre to >=128 or a single/standing titre of >=256 was

taken as diagnostic evidence of a L. pneumophila infection.
2. A single titre of >=128<256 was considered positive but not diagnostic of
L. pneumophila infection.
3. Single titres of less than 128 are considered negative.

Appendix 3. Comparison of results obtained with the RSIL IFAT using
FYSA and each commercial kit for sera from patients with proven L.
pneumophila serogroup 1 infection.
RSIL IFAT vs. Zeus Sgp 1-6 EIA
Zeus 1-6 EIA

Positive Negative Total
Positive 107 16 123
RSIL Negative 1 64 65
Total 108 80 188
Concordance = 91%
RSIL IFAT vs. Sigma Sgp 1-6 EIA
Sigma 1-6 EIA

Positive 96 27 123
Total 96 92 188
Concordance = 86%
RSIL IFAT vs. Zeus Sgp 1 IFAT
Zeus 1 IFAT
Positive 103 20 123
Total 103 85 188
Concordance = 89%
RSIL IFAT vs. MRL (Heat) Sgp 1 IFAT
MRL (Heat) 1 IFAT

Positive 83 40 123
Total 85 103 188

Concordance = 78%
Appendix 3 (cont’d). Comparison of results obtained with the RSIL IFAT using FYSA
and each commercial kit for sera from patients with proven L. pneumophila serogroup 1
infection.
RSIL IFAT vs. MRL (Formalin) Sgp 1 IFAT
MRL (Formalin) 1 IFAT

Positive 117 6 123
Total 117 71 188
Concordance = 97%
RSIL IFAT vs. Zeus Sgp 1-6 IFAT
Zeus 1-6 IFAT

Positive 104 19 123
Total 105 83 188
Concordance = 89%
RSIL IFAT vs. Gull Sgp 1-6 IFAT
Zeus EIA
Positive 115 8 123
Total 116 72 188
Concordance = 95%

Appendix 4. - Addresses of the distributors
Product Catalogue number Distributor
Zeus Legionella ELISA Test System 4ZS20002 Quest Biomedical Ltd.

Greville Court
Zeus Legionella IFAT System (Gp 1) 4ZS20125 1665 High Street
Knowle
Zeus Legionella IFAT System (Gp 1-6) 4ZS20127 West Midlands
B93 0LL
Tel :01564 775750

Fax: 01564 775759
Sigma Diagnostic Legionella EIA538-A Sigma-Aldrich Co. Ltd.

IgG/IgM/IgA (Manual) Fancy Road
Poole
EIA538-B Dorset
(Automated) BH12 4QH
Tel: 0800 373731

Fax: 0800 378785
MRL Diagnostic Legionella IFAT N/A The Binding Site Ltd.

P.O.Box 4073
Birmingham
B29 6AT
Tel: 0121-414 2000

Fax: 0121-472 6017
Gull Laboratories L.pneumophila IFAT 4873131 Launch Diagnostics Ltd.

(1-6) Ash House
Ash Road
Longfield
Kent
DA3 8JD
Tel: 01474 874426

Fax: 01474 872388

Elisa Legionella

Hochgeladen von

Dokumentinformationen

Originaltitel

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

Elisa Legionella

Hochgeladen von

Copyright:

Verfügbare Formate

Legionella pneumophila antibody assays

An evaluation of commercial kits and reagents for the

Respiratory and Systemic Infection Laboratory, CPHL Central Public

Synopsis of the study 3

RSIL Evaluation Report Legionella pneumophila antibody assays 2

In England and Wales serological confirmation of suspected

MATERIALS AND METHODS

RSIL Evaluation Report Legionella pneumophila antibody assays 3

Specificity/seroprevalence in a healthy population - Specificties

Specificity - The 219 sera from the 114 non-legionella pneumonia

Sensitivity - Estimates of sensitivity were made using sera obtained

Estimates of the Positive and Negative Predictive Values - PPV

RSIL Evaluation Report Legionella pneumophila antibody assays 4

The simplest option for the replacement of the CPHL L.pneumophila

The other L. pneumophila serogroup 1 monovalent IFAT antigen (Zeus)

Some laboratories may, for local organisational reasons, prefer to use

Cross-reactions due to campylobacter will continue to be a significant

Legionnaires’ disease (LD) is an uncommon form of pneumonia,

RSIL Evaluation Report Legionella pneumophila antibody assays 5

Culture and isolation of legionellae from respiratory specimens

Direct fluorescent antibody staining of respiratory specimens is a very

It is now well accepted that the use of enzyme immunoassays (EIAs)

In 1997 almost two-thirds of the patients with a confirmed diagnosis

RSIL Evaluation Report Legionella pneumophila antibody assays 6

To allow all PHLS laboratories to make an informed choice when

It should be noted that this study only sought to address the

RSIL Evaluation Report Legionella pneumophila antibody assays 7

Previous work has shown that all subclasses of Ig must be determined

RSIL Evaluation Report Legionella pneumophila antibody assays 8

Assay Name Maufacturer/Source UK distributor Product

RSIL Evaluation Report Legionella pneumophila antibody assays 9

Assay Name Maufacturer/Source UK distributor Product

RSIL Evaluation Report Legionella pneumophila antibody assays 10

Assay name Format Ig Class Specimen Antigens

RSIL Evaluation Report Legionella pneumophila antibody assays 11

Assay name Claims Negative Positive results Limitations and interpretation

RSIL Evaluation Report Legionella pneumophila antibody assays 12

A four-fold rise in titre

A standing or single titre

Single titres of less than

RSIL Evaluation Report Legionella pneumophila antibody assays 13

A four-fold rise in titre

A standing or single titre

Single titres of less than

A titre >=128 provides

RSIL Evaluation Report Legionella pneumophila antibody assays 14

An ‘outline’ evaluation protocol was circulated to the manufacturers/

Where provided, assays were performed and interpreted according to the

RSIL Evaluation Report Legionella pneumophila antibody assays 15

Pneumonia patients with infections of known aetiology

Sera were available from 114 non-legionella pneumonia patients with

Patients with falsely positive legionella serology

B. Patients with proven L. pneumophila infection

Proven L. pneumophila serogroup 1 infection

Proven L. pneumophila non-serogroup 1 infection

Six patients with infection proven by culture of L. pneumophila of

RSIL Evaluation Report Legionella pneumophila antibody assays 16

Age range Male Female Total

30-39 100 66 166

Table 6. Sera examined from patients with non-legionella pneumonia

Causative Agent No. of Patients No. of Sera

RSIL Evaluation Report Legionella pneumophila antibody assays 18

Serological findings Evidence of infection Totals