INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The reliability of protected specimen brushing (PSB) and bronchoalveolar lavage (BAL) depends largely on adequate sampling by a skilled bronchoscopist (1). Assessment of the quality of the recovered sample seems essential for the interpretation of microbiologic results.

In BAL, microscopic standards for sample quality have been proposed (1, 2). However, no such criteria exist for assessing PSB sample quality. PSB was introduced in our hospital as a tool for the diagnosis of nosocomial pneumonia in 1991. However, early sampling was rather unsatisfactory: cultures were almost always negative. Inadequate sampling technique was considered as a possible explanation for this, and the technique was adapted. The brush was advanced more peripherally and gently rotated. We assumed that by doing this, the brush would make contact with the bronchial mucosa, and that inflammatory and bronchial cells would be recovered in cases of local inflammation, and bronchial cells alone would be recovered in noninflammatory conditions.

To our knowledge, evaluation of cellular content has not been used as an indicator of adequate PSB sampling. In order to address the question of PSB sample quality, we retrospectively analyzed numbers of cells in cytospin preparations of PSB samples from a first sampling period and those from a second period after adaptation of the PSB technique.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patients

From February 1991 to August 1994, we evaluated all patients with suspected nosocomial pneumonia who were admitted to the medical intensive care unit and the department of pneumology of our institution. Nosocomial pneumonia was clinically suspected on the basis of a new or progressive pulmonary infiltrate on chest radiography at 48 h after admission, in association with fever (> 38° C), leucocytosis, and increased purulent bronchial secretions detected by visual inspection. The following data were recorded for each patient: age, sex, characteristics of infiltrate (localized/diffuse), immunosuppression status, underlying disease, mechanical ventilation, administration of antibiotics prior to the day of sampling, and outcome. From March 1992 to June 1994, PSB samples were obtained from 15 noninfected controls who required fiberoptic bronchoscopy for different indications (eight patients with lung cancer, three patients with chronic obstructive lung disease, one patient with a neurologic disorder, one patient with pulmonary infarction, and two patients with systemic autoimmune disease). The PSB sample serving as a control was obtained from the nonaffected side.

PSB Sampling Technique

As described by Marquette and colleagues (3), a single catheter (13127; Novatech, Vitrolles, France) with a double port on its proximal end was used. The brush rod is fitted with a plastic cap on its first port, thus ensuring airtightness. An air-filled syringe (10 ml) is connected to the second port, permitting ejection of the agar plug that is already in place and which occludes the distal end of the catheter.

After minimal use of topical anesthetic (xylocaine 1%), a fiberoptic bronchoscope (BFP 20; Olympus, Tokyo, Japan) was introduced and advanced under direct vision to a position next to the orifice of the third branch of the bronchus selected according to the location of the infiltrate on a chest radiograph. The catheter was passed through the inner channel of the fiberoptic bronchoscope. In the first study period, from February 1991 to February 1992, the catheter was advanced 2 to 3 cm beyond the distal end of the bronchoscope. During the second study period, from March 1992 to June 1994, the catheter was advanced into the lung tissue until resistance was encountered, and was then gently pulled back about 3 cm. The position of the bronchoscope was not controlled fluoroscopically. No other change in the technique was made. After the distal plug was ejected, the brush was advanced 2 to 3 cm beyond the tip of the catheter, rotated several times, and then retracted a few centimeters into the catheter. After removal from the fiberoptic bronchoscope, the distal portion of the catheter was cleaned with 70% ethanol, wiped dry, and transected distal to the brush with a sterile scalpel blade. The brush was then advanced beyond the sheath and vigorously vortexed in 1 ml of sterile Ringer's solution for at least 60 s to suspend all material from the brush. This sample was brought to the laboratory within 15 min.

The procedure was well tolerated by all patients except for one, who developed a hemorrhage in a subsegmental branch.

Laboratory Investigations

Quantitative culture was done with the "calibrated loop" method. The sample was vigorously vortexed before processing. Ten microliters of Ringer's lactate solution were inoculated onto Columbia agar with 5% sheep blood (Becton-Dickinson, Meylan, France), Centers for Disease Control anaerobe blood agar, McConkey's agar, and Sabouraud's dextrose agar with chloramphenicol. One hundred microliters of the sample were inoculated with a calibrated pipet on chocolate II agar (Becton-Dickinson) and buffered charcoal-yeast-extract agar (BCYE-). Plates were incubated at 36° C under adequate aerobic and anerobic conditions, and were evaluated for growth after 24 h, 48 h, and 5 d. Estimates of numbers of bacteria in the original fluid were made by colony counts of each morphotype, and were expressed in colony-forming units per milliliter (cfu/ml). A count of  10³ cfu/ml was used as the cutoff point for positive culture. Cultures were investigated according to standard procedures (4).

For direct microscopic analysis, two slides were prepared by cytocentrifugation with a Shandon Cytospin 2 (Southern Products, Funcorn, Cheshire, UK). With this technique, cells and microorganisms are centrifuged through a hole in a filter-paper strip, which absorbs the supernatant fluid of the specimen, and are deposited on a 6-mm-diameter circular area of a slide. For each slide, 140 µl of fluid were used. One May-Grünwald-Giemsa stain and one Gram stain were done on each specimen. Microscopic examination was done with a magnification of ×500 objective: ×50, eye piece: ×10) and oil immersion. Cell counts were made on five microscopic fields with high cellularity. Macrophages, polymorphonuclear neutrophils, lymphocytes, and columnar epithelial cells (bronchial cells) were considered in this count. Red blood cells were not considered. We scored the preparations according to three categories: less than ten cells per field, 10 to 20 cells per field, and more than 20 cells per field. A cell differential count was obtained by counting 100 cells in a May-Grünwald-Giemsa-stained preparation. Samples with more than 1% squamous epithelial cells and samples that showed the features of a peripheral blood smear were evaluated separately.

Statistical Analysis

Comparisons were made using Student's t test or chi-square test. Differences were considered significant at p < 0.05.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Samples from 86 patients with suspected nosocomial pneumonia were included in the study: 28 patients in the first period, from February 1991 to February 1992, and 58 patients in the second period, from March 1992 to June 1994. There was no statistical difference between the two groups in data recorded, such as sex/age distribution, characteristics of infiltrate (localized/diffuse), immunosuppression status, mechanical ventilation, underlying conditions, and outcome. In the first study period, 20 (71.4%) of the patients had received antibiotics before sampling. Of these 20, 19 (67.9%) received broad-spectrum antibiotics and one patient had received a glycopeptide. Mean duration of antibiotic treatment prior to sampling was 4.7 (± 3.7) d. In the second study period, 37 (63.7%) of the patients had received antibiotics before sampling. Of these 37, 31 (53.4%) had received broad-spectrum antibiotics, whereas six patients (10.3%) had received antibiotics with a spectrum limited to gram-positive organisms. Mean duration of antibiotic treatment prior to sampling in the second period was 3.9 (± 3.1) d. Although relatively more patients in the first than in the second period received antibiotics, the differences were not statistically significant. The results of cell counts in subsequent periods are summarized in Table 1. During the second period (March 1992 to June 1994), when the PSB sampling technique was adapted, only six samples (10.3%) contained less than 10 inflammatory or bronchial cells per field (objective: ×50), which contrasts strongly with the first period, in which 17 samples (60.7%) contained < 10 cells per field. Accordingly, the number of samples containing more than 20 cells per field increased from 28.6% to 74.2% from the first to the second study period.

In the second study period, considerably more positive quantitative cultures were obtained (26 of 58) than in the first period (four of 28). No positive cultures were obtained if less than 10 cells per field were present on direct examination. In both periods there was a significant difference (p = 0.001) in numbers of positive cultures from the category with < 10 cells/ field and the category with > 20 cells/field. Comparison of cell counts in the samples from the controls and from patients in the same study period is shown in Table 2. Controls and patients showed comparable results. All cultures of controls from PSB samples obtained from the nonaffected side were negative. The percentages of cell types recovered in samples from patients and controls were different. As shown in Table 3, the mean percentage of polymorphonuclear neutrophils was considerably lower in samples from controls, whereas bronchial cells were much more numerous. Of the samples of controls, 66.6% contained more than 90% bronchial cells, contrasting with only 12.2% of samples from patients.

Five samples contained more than 1% squamous epithelial cells. Of these five, four contained 1% to 2% squamous epithelial cells, whereas a fifth sample contained 27% squamous epithelial cells. Three of the five samples had positive culture results: two grew mixed oropharyngeal flora, and one grew Klebsiella pneumoniae and Enterobacter aerogenes.

Six samples showed the characteristics of a peripheral blood smear. One of these six samples yielded a positive quantitative culture.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Although a standardized methodology for PSB technique has been described (1, 5), no definition of a satisfactory PSB sample is proposed in the literature. Cellular content of cytospin preparations has not yet been validated as a criterion of PSB sample adequacy. In many studies, microscopic examination of PSB samples is not performed, and only microscopic examination of simultaneously collected BAL fluid is mentioned. However, the diagnostic value of Gram stain of PSB cytospin preparations has been demonstrated (6, 7). In a previous study, we showed that in the absence of polymorphonuclear neutrophils, a positive quantitative culture is unlikely, and that the presence of inflammatory cells containing intracellular organisms correlated strongly with a positive culture (7).

In the present study, we tried to demonstrate the correlation of cell numbers in cytospin preparations with the quality of PSB sampling. We assumed that in PSB the brush is either rotated in visible exudative secretions or gently rubbed on the bronchiolar mucosa. In either case, cells should be recovered. Inflammatory cells are present in exudative secretions. In the absence of exudate, friction of the brush on the mucosa should detach cells. Complete absence of cells therefore suggests that the brush rotated only in the lumen of a small bronchus or bronchiole, without making contact with the mucosa. In the absence of exudate, a brush turning in a vacant bronchiolar lumen can cause false-negative results, especially in early infection, in which inflammation is primarily present in the bronchiolar wall itself (8). We studied the number of inflammatory cells and bronchial cells in standardized cytospin preparations (volume: 140 µl) with a magnification of ×500 (objective: ×50). Our study shows that peripheral PSB sampling results in a higher number of cells in cytospin preparations. This was clearly shown in results of the two consecutive periods of the study. The two patient groups were very similar, and no obvious explanation, other than technical adaptation of PSB sampling, could be found for the higher cell content.

In our study, cultures of PSB samples with fewer than 10 cells per field were always negative. The observation that patients with PSB samples containing more cells more frequently yield positive cultures could be due to the greater ease of obtaining samples with a high cellular content from infected patients. We therefore examined PSB samples obtained from uninfected controls, and documented the possibility of obtaining specimens with high cellular content from these uninfected controls. Although the distribution of cell types was different in the patient and control groups, the total numbers of cells per field were comparable.

In neutropenic patients, a cellular inflammatory reaction will be less markedly expressed than usual, owing to their cytopenia. Although fewer inflammatory cells will be recovered from these patients, the numbers of cells recovered should not be lower than from uninfected controls. We did not study neutropenic patients, because BAL is preferred to PSB in this setting at our institution.

Some issues need further investigation. In accordance with accepted guidelines for BAL sample quality (1, 2, 5) we considered PSB samples that contained > 1% squamous epithelial cells to be contaminated by upper-airway sections. Sometimes, PSB samples are heavily contaminated with blood. Although hemorrhage is an occasional complication of brushing, its effect on culture results has not been reported. In our opinion these samples should be considered nonassessable in cases with negative cultures, since blood may have washed out or diluted the infectious material.

The inhomogeneous distribution of cells in the cytospin preparations in our study remains another point of concern. We counted at least five fields with high cellularity at a magnification of ×500 (objective: ×50), which is currently used in our laboratory. When a ×50 objective is not available, use of a ×40 objective is a possibility with a slightly higher cutoff (12 cells per field). Using a lower magnification could overcome part of the problem of inhomogeneity of the preparation. However, with a magnification of ×100 (objective: ×10), cells are very small and difficult to count. Magnification of ×250 (objective: ×25) and a threshold of 25 cells per field could be a better choice. However, one must keep in mind that the proposed thresholds should be considered guidelines rather than exact cutoff values. A "gold standard" that indicates a well-performed brushing in infected as well as in noninfected patients is currently unavailable, and an accurate cutoff cannot be determined. Numbers of cells around the proposed threshold should therefore be interpreted with caution.

Conclusion

Our study indicates that microscopic evaluation of PSB samples should always be included in the interpretation of culture results from these samples. Absence of cells probably represents inadequate sampling. Negative PSB cultures that yield cytospin preparations without cells should therefore be considered with caution, and resampling should be reconsidered. In our study, a threshold of less than 10 cells per field at a magnification of ×500 correlated with inadequate sampling.