Download 7.17.09 Cullen Legionella PCR

EBM 7/17/09, Legionnaires’ Disease
Citation:
Diederen BM, Kluytmans JA, Vandenbroucke-Grauls CM, and Peeters MF.
2008. Utility of Real-Time PCR for Diagnosis of Legionnaires' Disease in
Routine Clinical Practice. J Clin Microbiol 46:671-677.
Study Summary:
Objectives: To determine the value of PCR in the diagnosis of Legionnaires’ disease in routine clinical
practice.
Design: Retrospective data collection.
Setting: Data obtained from the Regional Public Health Laboratory, in St. Elisabeth Hospital, the
Netherlands. Specimens tested were those submitted for routine diagnosis of pneumonia from December
2002 to November 2005.
Patients: All hospitalized patients, had symptoms compatible with pneumonia, and showed radiological
signs of infiltration. To be included in the analysis, specimens had to have included PCR and either culture
for Legionella spp. or urinary antigen. Of the included patients (n=151), 61% were male with a mean age
of 57 years (range 23-74 years). Mean age of the female participants was 60 years (range 25-85 years).
Methods: A case of confirmed Legionella pneumonia (ie the gold standard) was defined according to the
European Working Group on Legionella Infections criteria. These consider an “LD-positive” patient as
one with a positive Legionella spp. cultured from a respiratory sample, or a positive Legionella urinary
antigen.
At the time of collection, respiratory samples were sent to the molecular biology lab for PCR, and
simultaneously cultured. PCR was targeted at two specific regions: 16S rRNA gene and the mip gene. All
positive PCR results were confirmed with a second PCR run on a second area of DNA isolated from the
respiratory sample – if negative, the sample as a whole was considered LD negative by PCR.
“Discrepant analysis” was performed on specimens for which PCR and the gold standard gave different
results. In an attempt to determine which one was correct, additional testing was performed to try and
confirm the PCR positivity, given the imperfection of the gold standard. A different PCR assay was used, a
different laboratory performed the same PCR (16S rRNA or mip) or sequencing analysis. If any of these
approaches confirmed a true positive, the specimen was considered “LD positive”.
Validity Criteria:
Are the results of this diagnostic study valid?
Was there an independent, blind
comparison with a reference (“gold”)
standard of diagnosis?
Was the diagnostic test evaluated in an
appropriate spectrum of patients (like
Yes, although the “gold standard” itself
is imperfect in this case. Reported
culture sensitivity varies from 41%-91%.
Urinary antigen testing has sensitivities
reported from 60-100%, but only detects
L. pneumophila serogroup 1.
Yes, for inpatient care.
those in whom it would be used in
practice)?
Was the reference standard applied
Yes
regardless of the diagnostic test result?
Was the test (or cluster of tests) validated No (only one group)
in a second, independent group of
patients?
Results:




Of the 151 specimens included in the study, 37 of these patients (25%) fulfilled the
EWGLI criteria as an “LD positive” case.
Of these 37:
 21 were culture positive
 35 were urinary Ag positive
 19 were culture and Ag positive
For PCR, two samples had inhibitors that could not be removed. Therefore, 149
patients were included in the final calculations.
 40 out of the 149 patients (27%) tested positive for Legionella based on PCR
diagnostic criteria
 36 were positive for L. pneumophila, and 4 for non-pneumophila Legionella
spp.
 37 tested positive by the 16S rRNA PCR (5 false positives and 5 false
negatives compared to GS)
 36 tested positive by mip PCR (2 false positives and 3 false negatives
compared to GS)
 33 were positive for both the mip and 16S rRNA PCRs
 3 were positive for mip only
Using discrepant analysis, 2/5 false positives in 16S rRNA group would actually be
true positives, and 2/2 in mip group. **My calculations are based on original data,
not the discrepant analysis reclassification.**
Blank 2x2 table for your own calculation purposes (mip pcr):
Totals
Totals
Sensitivity = a/(a+c) = 34/37 = 92%
Specificity = d/(b+d) = 110/112 = 98%
Likelihood ratio for a positive test result = LR+ = sens/(1-spec) = 92%/2% = 46
Likelihood ratio for a negative test result = LR - = (1-sens)/spec = 8%/98% = 0.08
Positive Predictive Value = a/(a+b) = 34/36 = 94%
Negative Predictive Value = d/(c+d) = 110/113 = 97%
Pre-test probability (prevalence using this target population – those hospitalized
and diagnosed with pna) = (a+c)/(a+b+c+d) = 37/149 = 25%
Pre-test odds = prevalence/(1-prevalence) = 25%/75% = 0.33
Post-test odds = pre-test odds  LR = 0.33 x 46 = 15.33
Post-test probability = post-test odds/(post-test odds +1) = 15.33/16.33 = 94%
Interpretation/Conclusions:
Can you apply this valid, important evidence about a diagnostic test in caring for
your patient?
Is the diagnostic test available, affordable,
accurate, and precise in your setting?
Can you generate a clinically sensible
estimate of your patient’s pre-test probability
(from personal experience, prevalence
statistics, practice databases, or primary
studies)?
 Are the study patients similar to your
own?
 Is it unlikely that the disease possibilities
or probabilities have changed since the
evidence was gathered?
Will the resulting post-test probabilities
affect your management and help your
patient?
 Could it move you across a testtreatment threshold?
 Would your patient be a willing partner
in carrying it out?
Would the consequences of the test help
your patient?
No, we do not carry Legionella PCR
testing in-house
Pre-test probability: Yes
(Responsible for 1-5% of CAP)
Study patients: All from the
Netherlands, where there may be
different strains of Legionella spp.
But overall, likely similar enough.
Changes: No
Would consider testing if high
suspicion with negative urinary
antigen, and particularly in cases of
possible outbreaks. (Head to head
analysis of urinary Ag vs. PCR was
relatively equivalent, but
combination of Ag and mip PCR
had some additive effect, p=0.7)
Likely – simple blood draw
Minimally. Empiric treatment for
CAP usually is either levofloxacin
or Zithromax, both of which treat
Legionella. However, if a patient
diagnosed with Legionella were not
improving, could alter treatment to
include a tetracycline or
erythromycin, which have been
shown to have better outcomes in
some studies.
Additional thoughts/Questions:





Possible there is a somewhat decreased specificity secondary to an imperfect gold
standard
All retrospective analysis
Discrepant analysis approach introduces bias by the selective use of confirmation
testing, and reclassification of “true” and “false” negatives and positives
Entire study population was biased to those in whom they felt the need to check for
Legionella rather than just treat empirically
Is there a significant number of non-penumophila Legionella spp causing clinically
significant pneumonia (only 1 case in 151 tested in this study)