| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Guest Access | Sign In via User Name/Password
|
|
|
|||||||
Guest Access | Sign In via User Name/Password |
|||||||||
Dr. Pesola is from the Divisions of Epidemiology and Biostatistics, Columbia School of Public Health.
Correspondence to: Gene R. Pesola, MD, MPH, Divisions of Epidemiology and Biostatistics, The Mailman School of Public Health, Columbia University at 168th St, 600 West 168th St, Ph18, New York, NY 10032
Pneumonia, the leading cause of death in 1900, is currently the sixth-leading cause of death when combined with influenza,1 and necessitates hospitalization in > 485,000 noninstitutionalized adults in the United States each year, with approximately 43,000 deaths.2 3 Estimates suggest that 77% of pneumonia occurrences are treated in the outpatient setting4 ; this would extrapolate to > 2 million pneumonia occurrences each year in the nonpediatric population. Approximately one sixth to one fifth of hospitalized adults are believed to have streptococcal pneumonia (pneumococcus), based on the assumption of a 25 to 30% bacteremic rate.3 5 Etiologic estimation of the outpatient adult population with pneumonia is difficult due to the lack of data, but a few studies5 6 7 suggest that 9% or more are pneumococcal in origin. Overall evidence suggests that despite a slowly declining estimated incidence of > 50% of pneumonias in the 1960s, at least in hospitalized patients,8 9 to < 25% of pneumonias currently based on outpatients and inpatients with pneumonia, the pneumococcus is presently still the number-one cause of adult pneumonia. Therefore, any diagnostic test that can improve the diagnosis of pneumonia in general, and in particular the most common pneumonia, would be of particular benefit to the primary-care physician who is meeting the patient at the initial point of care.
In this issue of CHEST (see page 243), Dominguez et al
describe an immunochromatographic assay (ICT) that detects urinary
C-polysaccharide antigen, a cell wall antigen believed to be specific
for all pneumococcal serotypes, as well as for Streptococcus
oralis and Streptococcus mitis.10
The test
takes 
15 min to perform and involves inserting a dipstick into a
random urine sample of a patient presenting with pneumonia, and then
reading the result from the test container. The rationale for the test
is that soluble or particulate microbial antigens from streptococcal
pneumonia are excreted in a concentrated state in the urine, making the
urine a superior site to detect antigen.11
When using a
positive blood culture result and/or counterimmunoelectrophoresis
detection of urinary polysaccharide capsular antigen using pneumococcal
omniserum (contains antibodies to all 84 serotypes of the pneumococcus)
as the criterion standard for the diagnosis of pneumococcal pneumonia,
the sensitivity of the test was 80% (41 of 51). The specificity of the
test was 97% (69 of 71) when comparing other diagnosed pneumonias,
other nonstreptococcal bacteremic conditions without pneumonia, and
urinary tract infections. One problem with the test as noted by the
authors is that the color intensity of the reaction for a positive
diagnosis is read as weakly positive, medium positive, or very
intensely positive as compared with the positive control line. Given
the well-known decreased performance of a test in the community
relative to a study site (ie, efficacy vs effectiveness),
the authors suggest that it may be more prudent to consider positive
results to be the medium and very intensely positive reactions only.
This would change the sensitivity to 74.5% (38 of 51) and the
specificity to 100% (71 of 71), still giving a solid performance for
the urinary ICT. As noted above, a cross-reaction of the urinary ICT
can occur with S oralis and S mitis, both normal
flora of the oropharynx and both known to cause endocarditis and not
pneumonia.12
Therefore, the clinical presentation of
pneumonia should not result in any cross-reactivity with these
antigens.
One drawback with what appears initially to be a very good test is revealed on analysis of their second group of patients (summarized in Table 2). This group is given a presumptive diagnosis of pneumococcal pneumonia based on a Gram’s stain consistent with the diagnosis and/or a positive sputum culture finding and/or latex agglutination positive in sputum samples for streptococcal pneumonia. Using this sample of patients, the urinary ICT is positive in only 7 of 16 patients, for a sensitivity of 44% as noted by the authors. Adding the further assumption that only moderate or intensely positive urinary ICTs would be reliably read when generalized to the clinical nonresearch or community setting, then only 2 of 16 streptococcal pneumonia patients would be identified by the new test, for a sensitivity of 12.5%. Therefore, the overall sensitivity of the urinary ICT for the detection of pneumococcal pneumonia is actually unknown and is probably somewhere in between the two estimates of 12.5% and 74.5%. Spectrum of disease bias occurs when a test sensitivity is reported in patients with only one extreme of disease and is minimized here by having two groups of patients with different extremes of disease. This bias is minimized for specificity as well, by reporting the test in patients with a broad range of nonpneumococcal disease. It should be noted that since the specificity of the test is very good, an unequivocally positive test result (medium to strongly positive) is helpful in suggesting the diagnosis of pneumococcal pneumonia and should direct primary therapy. A negative urinary ICT result is not helpful.
The ability to make a point of contact or immediate diagnosis of pneumococcal pneumonia in the past has been relegated to the sputum Gram’s stain with reports of a sensitivity up to 60% and specificity of 80%.13 Unfortunately, the technique is very operator dependent14 ; many patients cannot produce sputum; going through the steps necessary to obtain and evaluate a Gram’s stain sputum culture is more time-consuming than people will allow; and the result is still not definitive when positive, since the specificity does not approach 100%. Therefore, in general, there has been a tremendous decline in the use of the sputum Gram’s stain in clinical medicine. The urinary ICT, on the other hand, only requires an easily obtainable urine sample, and if the results are unequivocally positive in a patient with pneumonia, the diagnosis will be hard to refute. In addition, training of personnel should be easy, in that practice with a positive and negative control should be all that is necessary. The only rate-limiting factor in its use will be cost. In the hospital setting, the probability of a positive test result may be as high as 15% (0.745 x 0.2) if the proportion of pneumococcal pneumonia among all pneumonia presenting to this setting is one in five. In the office setting, the probability of a positive test result may be as high as 7.5% if the proportion of ambulatory pneumococcal pneumonia is about 10% of presenting pneumonias. If the test is relatively inexpensive or in some way easy to pay for, a positive test result could direct definitive therapy, while a negative test result would be of no diagnostic benefit. Only time will determine whether the test will be utilized routinely relative to its cost, since there will be a lot of negative results.
Use of the ICT raises a number of issues:
1. In the ambulatory setting, in a patient without a chest radiograph or sputum and with a clinical presentation consistent with pneumonia, could a positive urinary ICT result be caused by a viral syndrome with simple colonization of the oropharynx? Could the same scenario with green/yellow sputum be simple pneumococcal bronchitis?
2. Could the simple step of concentrating the urine increase the sensitivity of the ICT and increase diagnostic yield? In addition, is that what actually occurred in the article by Dominguez et al, as noted in the "Discussion" section? If so, this would imply that the sensitivity of the test as reported is higher than what would occur in actual practice.
3. Many studies of the etiology of community-acquired pneumonia are unable to determine up to 50% of the final diagnosis. Will this test increase the diagnostic yield of future studies in detecting the pneumococcus among the unknowns?
4. Up to 10%15 of community-acquired pneumonia cases have more than one organism. If the urinary ICT result is positive, would it be reasonable to focus therapy on just the pneumococcus, or should a more broad-spectrum therapy be considered? Most likely, the severity of disease will guide therapy in this instance.
It appears that Dominquez et al have taken a significant step forward in the definitive diagnosis of pneumococcal pneumonia. The strength of their study appears to lie in the rapidity of making a diagnosis as well as the incredible specificity of the test (ie, a minimal number of false-positive results), implying that when a test result is positive the diagnosis is correct. The weakness of their study lies in the unknown sensitivity of their test, which will require larger studies using a broad spectrum of disease to further tease out a closer estimate of the actual sensitivity. Further similar studies are needed from different investigators at different centers to verify that the results they obtained are indeed correct. If this is the case, more focused, cost-effective therapy can be utilized in the treatment of this common pneumonia and studies evaluating pneumonia etiology should be enhanced. Point-of-contact treatment of streptococcal pneumonia per se will not change with the test, since this organism is always covered, although therapy might be more focused depending on results and clinical circumstances.
References
-hemolytic streptococci possessing a C-polysaccharide phosphorylcholine-containing antigen. Infect Immun 61,3076-3077This article has been cited by other articles:
|
|
 |
|
  N. Sopena, M. Sabria, and the Neunos 2000 Study Group Multicenter Study of Hospital-Acquired Pneumonia in Non-ICU Patients Chest, January 1, 2005; 127(1): 213 - 219. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. Dominguez, N. Gali, S. Blanco, P. Pedroso, C. Prat, L. Matas, V. Ausina, and G. R. Pesola Urinary Antigen Test for Pneumococcal Pneumonia Chest, November 1, 2001; 120(5): 1748 - 1750. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
