| 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 |
|||||||||
* From the Departments of Medical Microbiology (Drs. Jacobs, De Brauwer, and Ms. Cornelissen) and Pulmonology (Dr. Drent), University Hospital Maastricht, Maastricht, The Netherlands.
Correspondence to: Jan A. Jacobs, MD, PhD, Department of Medical Microbiology, University Hospital of Maastricht, PO Box 5800, 6202 AZ Maastricht, The Netherlands; e-mail: JJA{at}ms-azm-3.azm.nl
 |
Abstract |
|---|
 TOP Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Design: Prospective. The relative PMN counts (obtained by conventional microscopy and expressed as a percentage of a 500 cell count) of consecutive BAL fluid samples were compared with the corresponding LE categories as read with a urine chemistry reader. LE categories were graded as follows: negative, trace, +, + +, and + + +.
Results: A total of 153 BAL fluid samples were included. The mean PMN counts of the negative LE category (4.1 ± 4.3%; n = 43) and the + + + category (81.8 ± 16.3%; n = 37) differed significantly from each other and from the mean PMN counts of the other categories. Within the trace, +, and + + categories, a considerable overlap of PMN counts was noted. Assignment of a BAL fluid to the negative LE category consistently predicted a PMN count < 20%. At a threshold value of 50% PMNs, the + + + LE category predicted the BAL fluid samples to the correct group (PMNs > 50% vs < 50%) with a sensitivity of 70.8% and a specificity of 97.1%.
Conclusions: The reagent strips proved to be useful as a rapid test for semiquantitative measurement of the relative PMN counts in BAL fluid. However, the low predictive value for the exclusion of a high PMN count may limit their application.
Key Words: BAL • leukocyte esterase • polymorphonuclear neutrophils • reagent strips • ventilator-associated pneumonia
|
Introduction |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
Materials and Methods |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Sampling Technique
A fiberoptic bronchoscope (Pentax FB-15 h/FB-15X; Pentax
Medicals; Tokyo, Japan) was introduced through a special adapter
(Swivel Connector; Gibeck Respiration; Upplands Väsby, Sweden)
and "wedged" into the affected segmental or subsegmental bronchus.
The fluid was instilled into the subsegment through the biopsy channel
of the bronchoscope in four 50-mL aliquots of sterile saline solution
(0.9% NaCl, room temperature) and immediately aspirated and recovered.
The BAL fluid samples were transported to the laboratory within 15 min
after collection and analyzed within 1 h on arrival in the
laboratory.
Cytologic Processing of BAL Fluid Specimens
The volume of the recovered BAL fluid and its macroscopic
appearance were recorded. The first fraction, representing the
bronchial fraction, was separated for mycobacterial culture and the
remaining fractions were pooled in conical 50-mL nonadhesive
polypropylene tubes (Product No. 227.261; Greiner; Alphen aan de Rijn,
The Netherlands). The total cell count was performed in a
Fuchs-Rosenthal hemocytometer chamber. All nucleated cells were
counted, and the average value of two successive counts was considered.
Cytocentrifugation was done with the Cytospin 3 apparatus (Shandon
Scientific; Astmoor, England) as previously described.7
The slides were allowed to air dry, stained with the May-Grünwald
Giemsa dyes, and subsequently sealed with a cover glass by means of a
xylene-free mountant (Histomount; Shandon Scientific). Differential
cell counts were made by examining 500 nucleated cells, excluding
squamous and bronchial epithelial cells. Cells were counted in a
circular pattern around the center of the cytocentrifuge
spot,8
and the differential cell counts including the PMNs
were expressed as a percentage of a 500-cell aliquot.
Rejection Criteria
BAL fluid samples were excluded if the recovery was < 20 mL or
if the total cell count was < 60,000 cells/mL, and if the
cytocentrifuged preparations showed excessive amounts of RBCs,
intercellular debris, or damaged cells precluding adequate recognition
of different cell types.7
9
Reagent Strips
Multistix 7 reagent strips were used for semiquantitative
assessment of the BAL fluid LE activity. Multistix 7 is a seven-patch
test reagent strip that was designed to test urine for glucose,
ketones, blood, pH, protein, nitrite, and leukocytes. The reagent
strips were read instrumentally, using the Clinitek 50 Urine Chemistry
Analyzer (Bayer Corporation). The performance of the reagent strips was
checked against the Chek-stix positive and negative controls (Bayer
Corporation) at any time a new package was begun. The procedures were
performed as described in the package insert and in the Clinitek 50
operating manual.
Reagent strips were removed from the bottle just immediately before they were used for testing. They were dipped directly and briefly into well-mixed pooled BAL fluid specimens. If the BAL fluid specimen depth in the conical tubes was < 5 cm, the specimen was poured into a 10-mL tube. While removing the strips, the edge of the reagent strip was dragged against the rim of the BAL fluid container to remove excess fluid. At that time, the "Start" key of the Clinitek 50 reader was pressed. Within the provided 5-s to 10-s delay, the reagent strip was blotted by gently touching the edge to a paper towel. Subsequently, the reagent strip was placed into the trough of the test strip table of the Clinitek 50 reader and slid along the table until it touched the end of the trough. Readings were performed automatically, and test results were printed by the Clinitek 50 reader. The presence of PMNs was as graded as follows: negative, trace, +, + +, and + + +.
To compare the instrumental readings with visual readings, a number of reagent strips were also read against the standards provided on the bottle label, at 2 min after dipping, according to the Multistix 7 package insert. Furthermore, the effect of increasing reaction times for the reagent strips was studied on four BAL fluid specimens that were diluted with 0.9% NaCl until they were read as + by the Clinitek 50 reader. Reagent strips were subsequently entered in the Clinitek 50 reader at increasing delays of 30, 60, and 120 s after dipping.
Statistical Evaluation
For the different LE categories, pair-wise comparisons of the
mean PMN counts were assessed for significance by the one-way analysis
of variance post hoc test for multiple comparisons, with
Bonferroni’s modification
|
Results |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
In Figure 1 , the semiquantitative LE categories are plotted against the corresponding numbers of BAL fluid PMNs. As can be seen in Figure 1 , a considerable overlap of individual PMN counts was noted among the LE categories of trace, +, and + +. The respective mean PMN counts of these categories did not differ significantly (mean ± SD PMN counts, 20.2 ± 22.8%, 27.1 ± 21.5%, and 32.8 ± 21.4% respectively). However, the mean PMN count of the negative LE category (4.1 ± 4.3%; n = 43) differed significantly from the mean PMN counts of the trace category (p = 0.003) and the + + and + + + categories (p < 0.001). Assignment of a BAL fluid to the negative LE category consistently predicted a PMN count of < 20%. Conversely, the mean PMN count of the + + + LE category (81.8 ± 16.3%; n = 37) largely exceeded the mean PMN counts of any other category (p < 0.001). At a threshold of 50% PMNs, the + + + LE category predicted the BAL fluid samples to the correct group (PMNs > 50% vs < 50%) with a sensitivity of 70.8% and a specificity of 97.1%. Given the prevalence of 31.4% for a PMN count > 50% in the present study, the + + + LE category had a positive predictive value of 91.9% and a negative predictive value of 87.9%.
|
The LE categories that were visually read on 40 BAL fluid samples corresponded well with those generated by the Clinitek 50 reader, except for an occasional discrepancy in the trace or + categories. By contrast, accurate timing proved to be critical for obtaining reliable results, as the readings of the BAL fluid samples shifted from the LE + category to the + + + category when the delay of entering the reagent strip into the Clinitek 50 reader reached 120 s. Likewise, visual readings at < 2 min or > 2 min after dipping caused false results into a too-low or a too-high LE category, respectively.
|
Discussion |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
One might argue that the present study population was heterogeneous, comprising BAL fluid samples obtained from patients receiving ventilation, as well as from patients who were immunosuppressed or suffering from interstitial lung diseases. However, the aim of this study was the validation of the Multistix 7 reagent strip as a prediction for the BAL fluid PMN count, irrespective of the conditions affecting these counts. Elevations of the BAL fluid PMN count may occur in several clinical conditions, and the relative numbers of PMNs are associated with disease severity in disorders such as Pneumocystis carinii pneumonia,14 sarcoidosis,4 ARDS,15 and VAP.6 In diagnostic practice, rapid estimation of the BAL fluid PMN count will be of most diagnostic gain in the intensive care setting, ie, when VAP and ARDS are suspected. BAL fluid specimens obtained in this setting are generally processed in the microbiological laboratory, where cytologic expertise is not always available. Furthermore, in cases of VAP and ARDS, high BAL fluid PMN counts are to be expected,15 16 and the need for prompt administration of antibiotics in case of VAP calls for a rapid test.17 In addition, in our experience, up to 40% of BAL fluid specimens sampled in the ICU arrive at the laboratory during weekends or off-hours. For these reasons, a rapid and reliable screening test for the prediction or the exclusion of high BAL fluid PMN counts is mandatory.
For a screening test to be adopted in the diagnostic laboratory, both a high sensitivity and a high specificity are required.18 Considering the LE + + + category in its ability to predict a > 50% PMN count, it is clear that the Multistix 7 reagent strip fulfills only the specificity criterion. This will be of particular importance when the reagent strip will be used in a setting with a high prevalence of elevated BAL fluid PMNs, such as in the ICU. In that case, the positive predictive value, or the chance that a BAL fluid assigned to the + + + LE category actually will display a > 50% PMN count, will increase. However, the negative predictive value, or the chance that a BAL fluid assigned to one of the other categories (except the negative category) actually has PMN counts < 50%, will decrease. By consequence, the low predictive value in the exclusion of high PMN counts will be the limiting factor of the reagent strips in this circumstance.
Given the concern on the moderate sensitivity, we looked for explanations for the false-negative events with respect to the 50% PMN threshold. For urinary samples, factors that may contribute to false-negative results include high protein levels,10 and, according to the Multistix 7 package insert, elevated glucose concentrations, the presence of cephalosporins or tetracyclines in high concentrations, and the presence of colored substances such as nitrofurantoin. In the present study, elevated protein levels were found in three of four false-negative specimens of the trace category, but they were equally observed among the correctly predicted samples. Although we presently did not score the use of antibiotics prior to bronchoscopy, we do not expect these agents to be responsible for false-negative readings, as BAL fluid generally represents a 1:10 to 1:100 dilution of the alveolar epithelial lining fluid.19 Furthermore, no readings that were obscured by heavily bloodstained specimens were presently found, and we repeatedly observed that high eosinophil counts (> 40%) did not interfere with the reagent strip readings. For increase of the sensitivity of the Multistix 7 reagent strip in the prediction of high PMN counts, further investigation of the possible causes of false-negative readings is required. In addition, we suggest further technical developments of the LE reagent strip area to include an extension of the reading scale, in an attempt to differentiate and grade the PMN counts of the trace, +, + + categories.
In the present study, we did not intend to evaluate the other reagent areas of the Multistix 7 reagent strip. In retrospect, most of these tests proved not to be reliable or clinically useful, except for the pH and the protein tests. The nitrite test, which points to the presence of Gram-negative bacteria, may not be used in this setting because dietary nitrate is not present in BAL fluid specimens. For purposes of the present study, the reagent strip could be simplified to the area of interest. One might equally consider the addition of areas for estimation of the other BAL fluid cell populations, eg, an area detecting nonspecific esterase enzyme activity, which in BAL fluid is confined to the alveolar macrophage population.20
We further found the Multistix 7 reagent strips convenient to use. The readings were easy to perform, although accurate timing proved to be essential for obtaining reliable results. Visual readings that were consistent with the instrumental readings and a long package shelf life made the occasional use of the strips easier. The Clinitek 50 reader was a reliable apparatus that required little maintenance and was easy to calibrate.
In conclusion, the Multistix 7 reagent strip designed for urinalysis may be used as a rapid test for semiquantitative estimation of the relative PMN count in BAL fluid samples, in those conditions in which facilities for cytologic examination are not available. The LE categories generated by the reagent strip may distinguish BAL fluid samples displaying elevated (> 50%) PMN count with a high specificity but at the cost of a low sensitivity. Further study on the false-negative events and refinements of the reading scale are desired for optimal use of this reagent strip in the present setting.
|
Footnotes |
|---|
Received for publication December 15, 1999. Accepted for publication May 8, 2000.
|
References |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
