Comparison of Bronchoalveolar Lavage Cytospins and Smears in Dogs and Cats
Differences in the cytological interpretation of bronchoalveolar lavage fluid (BALF) after cytospin preparation (CP) or manual smearing of pelleted cells preparation (MSP) were investigated in client-owned dogs and cats with inflammatory or infectious lower respiratory disease. Bronchoalveolar lavage fluid from healthy cats was also examined. With MSP, cell lysis was more frequently observed, and cellular distribution was more heterogeneous throughout the slide. When samples from healthy and diseased animals were considered together, a significantly greater percentage of neutrophils was seen on CP than on MSP slides (P<0.002). Cytospin preparations were considered of better quality in all individual comparisons. Cytospin preparation is advised in the evaluation of BALF with low total cell count. When only MSPs are evaluated, clinicians should be aware that differential neutrophil counts may underestimate the counts found on CP slides.
Introduction
Bronchoalveolar lavage (BAL) is a relatively safe diagnostic procedure that is used to evaluate conditions of the lower respiratory tract, to investigate the pulmonary response to spontaneously arising and induced diseases in many animal species, and to monitor progression of disease and response to therapy.1–3 In dogs and cats, cytological evaluation and microbial culture of BAL fluid (BALF) can provide important diagnostic information in inflammatory and neoplastic pulmonary diseases.4–7 Fluid recovered from the lungs can be analyzed using cytological, biochemical, microbiological, and immunological techniques.
Important cytological criteria of evaluation in the analysis of BALF are the number and types of cells recovered. Considerable variation has been reported in total and differential cell counts from BALF samples collected from dogs and cats.6 Most of the variation probably reflects changes in sampling technique, because a strict standardization of the procedure is lacking.6 During BAL, the results can be modified by the volume of fluid used, the time between infusion of lavage fluid into a bronchus and aspiration of the fluid, and the suction pressure during aspiration of fluid.8
Following fluid collection, the method of processing the BALF can significantly affect interpretation of cytological preparations in people and horses, although information on such effects in dogs and cats is limited.9–14 Hence, modifications in some parameters of BALF processing (such as time between collection and processing of the fluid, temperature or use of fixative, speed, acceleration rate and duration of cytocentrifugation, resuspension techniques, and the type of stain used) can modify the results of cytological examination.9–15
One study has shown no significant difference between the percentage of recovery of instilled fluid when the animal is placed into either right lateral or dorsal recumbency during the procedure. However, in the same investigation, smears prepared 24 hours after collection had a significant decrease in the percentage of neutrophils and eosinophils and a consequent increase in the percentage of macrophages, when compared with smears prepared within 3 hours of lavage.14 In addition, the method of cytological examination (for example, the total number of cells counted and the area of the slide used for counting) can also influence differential cell counting.12,16 However, the most important factor affecting the number of cells counted is probably the method of slide preparation. Reported slide preparation methods include cytocentrifugation, pellet smears, filter, or glass cover preparations.17–20
In small animal practice, a cytocentrifuge is not often available; however, preparing smears from pelleted cells after centrifugation is generally possible. Smear preparations have been compared to cytocentrifuged preparations in humans and horses, and although some differences in differential cell counts were noted, smears were proposed to offer a practical alternative for practitioners.19,20
The aim of the present study was to investigate the presence of significant differences in the interpretation of cytological preparations of BALF following either cytocentrifugation (cytospin preparation [CP]) or manual smearing of pelleted cells obtained after centrifugation (MSP). Samples of BALF were derived from healthy cats and from dogs and cats with nonbacterial inflammatory or bacterial conditions of the lower respiratory tract.
Materials and Methods
Animals
Bronchoalveolar lavage was performed in nine healthy, experimental, domestic European cats (all aged 2 years) as well as nine dogs (aged 5 months to 12 years; mean ± standard deviation [SD] 4.46±3.44 years) and 10 cats (aged 1 to 6 years; mean±SD 3.7±1.82 years) of various breeds with nonbacterial inflammatory or bacterial lower respiratory tract disease [Table 1]. Experimental cats were housed and cared for according to the national guidelines and the principles advised by the European Council for the Care of Laboratory Animals. The study was approved by the Animal Ethical Committee of the University of Liège.
All diseased dogs and cats were client-owned and were presented to the Liège University Small Animal Teaching Hospital. Diagnoses were primarily based on a combination of clinical, radiographical, hematological, bronchoscopical, microbiological, and cytological findings, according to criteria described in the literature [Table 1].4,21–24
Four of the dogs in the study had acute bacterial bronchopneumonia. These dogs had an alveolar pattern on thoracic radiographs and significant aerobic bacterial growth on culture of BALF. The clinical histories and physical findings in these dogs supported this diagnosis. Five dogs had chronic bronchitis. These dogs were in good general condition with the exception of a chronic cough. Radiography revealed a bronchial and/or interstitial pulmonary pattern, and no significant aerobic bacterial growth was seen on culture of BALF. One of these five dogs had eosinophilic bronchopneumopathy as defined by the presence of numerous eosinophils in the BALF and eosinophilic infiltration of the bronchial mucosa on biopsy.
Two cats had bacterial disease. One of these cats had confirmed infection with Mycobacterium avium, and another had an alveolar pattern on thoracic radiography and significant aerobic bacterial growth on culture of BALF. The 10 remaining cats had nonbacterial inflammatory disease characterized by either an eosinophilic or mixed eosinophilic and neutrophilic bronchitis and insignificant aerobic bacterial growth on culture of BALF.
Collection of BALF
Cats were sedated with medetomidine (100 μg/kg intravenously [IV] or intramuscularly [IM]), and dogs were sedated with either medetomidine (100 μg/kg IV or IM) or acepromazine-buprenorphine (0.03 mg/kg and 15 μg/kg IV or IM, respectively). Prior to induction with propofol (Diprivan ≤3 mg/kg), a 5-minute preoxygenation period was used, and oxygen saturation was controlled during the procedure using pulse oximetry continuous monitoring. Oxygen was instilled through the endoscopic biopsy channel but not during the lavage procedure. When the oxygen saturation fell to <80 (approximately), the endoscope was quickly withdrawn and the animal was eventually intubated. Oxygen was placed in front of the larynx or delivered through the endotracheal tube.
Animals were placed in sternal recumbency. Bronchoalveolar lavage was performed using a flexible pediatric endoscope (video bronchoscope ONIS EB-410S; 0.6 cm in diameter). In dogs, 20 mL of lavage fluid was instilled twice into the left main bronchus and once into the right main bronchus; in cats, aliquots of 5 to 10 mL were used. When the disease was localized, lobes for lavage were selected based on radiographical and gross bronchoscopical lesions. Each aliquot of preheated (37°C), sterile saline (sodium chloride 0.9%) was instilled into the lung through a three-way stopcock into the biopsy channel by syringe. After 3 seconds, it was withdrawn by low-power aspiration pump and collected into a sterile glass trap. The volume of the recovered fluid was quantified. A 2-mL aliquot was placed into an ethylenediaminetetraacetic acid (EDTA) tube for determination of total cell count by a hemocytometer. Another 2-mL aliquot was placed into a sterile tube for quantitative bacteriology and mycology. The remainder of the fluid was processed as described below.
Processing of BALF
The remaining nonfiltered BALF was divided in two aliquots. To decrease cell lysis, one aliquot was immediately centrifuged at 4°C for 15 minutes (2500 rpm, 915 g), and smears were manually prepared from the pelleted cells by streaking the cell suspension onto a microscope slide with the aid of a “spreader” slide used at a 45° angle. The second aliquot was stored at 4°C until centrifuged within 6 hours in a cytospina at room temperature for 4 minutes (1400 rpm, 197 g). Pelleted smears were stained in the clinic with Diff-Quick, whereas CP slides were stained in a laboratory with May Grünwald-Giemsa.
Evaluation of Slides
Both series of slides (CP versus MSP) were independently and blindly evaluated by two different observers (Dehard and Bernaerts) using a Nikon Eclipse E800 microscope. Differential cell counts (neutrophils, lymphocytes, macrophages, eosinophils, and basophils) were established by counting a total of 100 cells at high power (×63 objective). Epithelial cells and isolated nuclei were not counted.
Lysis of cells (evaluated by the relative proportion of isolated nuclei), presence of ciliated cells and anthracosis (black granules in the cytoplasm of macrophages), conservation of cilia in ciliated cells, and presence or absence of mucus (pink, filamentous background material) were qualitatively assessed on each slide.
Two different types of macrophages were observed. Undifferentiated alveolar macrophages predominated and were defined as being relatively large (10 to 15 μm), round to oval cells with eccentric round to oval nuclei and abundant granular cytoplasm. By contrast, activated alveolar macrophages were larger, and their cytoplasm was more abundant and frequently included numerous cytoplasmic vacuoles [Figure 1].
Neutrophils were morphologically similar to those found in peripheral blood [Figure 2]. Eosinophils were identified on the basis of the presence of variably sized, distinct eosinophilic cytoplasmic granules [Figure 3].
Two different types of lymphocytes were also observed. Small lymphocytes had rounded nuclei with dense nuclear chromatin and a scant rim of poorly stained cytoplasm [Figure 4]. Reactive (activated) lymphocytes were larger with more abundant, more intensely stained cytoplasm and eccentrically placed nuclei [Figure 4]. Ciliated epithelial cells were low to tall columnar with basally located hyper-chromatic nuclei [Figure 5].
Statistical Analysis
Statistical Analysis System (SAS) was used for statistical analysis.25 Total cell counts in dogs with bacterial and non-bacterial inflammatory respiratory disease were compared using a t-test. Similarly, t-test was also used to compare total cell counts in the three categories of cats.
Effects of observer and type of cytological preparation (CP or MSP) on the differential cell counts were evaluated for each cell type (macrophages, neutrophils, lymphocytes, and eosinophils) by means of analysis of variance (ANOVA) using generalized linear model procedures. Differences were considered significant when P<0.05.
Results
Quality of Slides
On CPs, cellular distribution was more homogeneous, cell lysis was less frequently observed, and more cells were available for counting than on MSPs. On MSPs, cell distribution was more heterogeneous, and cells were preferentially located at the leading edge of the smear.
The MSPs from two cats with nonbacterial inflammatory disease were unavailable for assessment because of the poor quality of cells. By contrast, all CPs were of good quality and could be evaluated in all cases.
Total Cell Count
Total cell count was increased in nonbacterial inflammatory and bacterial disorders in cats [Table 2]. Most differences were significant despite a large variability among all diseased animals.
Differential Cell Counts
In healthy and diseased cats, macrophages were the predominant cell type present, followed by neutrophils. Sparse lymphocytes were present, especially in healthy cats, and these were primarily small lymphocytes [Table 2]. The highest neutrophil and lowest macrophage counts were observed in those cats with bacterial disease [Table 2]. Although eosinophil counts were highest in cats with non-bacterial inflammatory diseases, no significant differences were seen in eosinophil counts among healthy animals or animals with bacterial or nonbacterial inflammatory diseases. In one single case (a cat with nonbacterial inflammatory disease), eosinophils were detected only on the MSP. Red blood cells were present in only two samples—from one dog and one cat with severe, nonbacterial inflammatory lesions of the tracheal or bronchial mucosa, in which the BALF macroscopically appeared to contain some blood.
The percentage of macrophages was lower and the percentage of neutrophils was higher in CP compared with MSP, although this overall effect was significant for neutrophils only (P=0.002). When CP and MSP data were compared for each individual category separately (i.e., normal cat, cat with bacterial disease, cat with nonbacterial inflammatory disease, dog with bacterial disease, and dog with nonbacterial inflammatory disease), the only significant difference was in the percentage of neutrophils in dogs with bacterial disease (P=0.0215). No significant observer effect was found (P≤0.05).
Ciliated Cells, Mucus, and Anthracosis
Ciliated cells were present in higher numbers on slides from dogs with respiratory disease compared with cats. This observation held for both types of cytological preparations. The cilia of these epithelia appeared more damaged on cells within MSP than on those within CP. Mucus was a frequent finding in both types of slides [Figure 1].
Discussion
The present study has investigated differences in the interpretation of cytological preparations obtained from BALF collected from healthy and diseased cats and diseased dogs after CP or after MSP following centrifugation.
A large number of parameters influence data obtained from analysis of BALF, including the BAL procedure itself, the period between collection and processing of the fluid, and techniques of processing and examining the slide.8–12,14–16,26–30 The total volume of BALF instilled has been shown to influence the total cell count and differential cell count, as well as the concentration of cells recovered.29 Lavage with larger volumes yielded lower total cell counts and lower concentrations of neutrophils, lymphocytes, and macrophages.29 Some authors recommend a total BALF volume of 2 mL/kg,31 while others recommend a constant volume of BALF, as used in the present study.1,21,22
Immediate withdrawal of instilled fluid is necessary to obtain an adequate cell yield and minimize loss of cells. Delay in retrieval will result in dissemination and absorption of the fluid by the lung parenchyma and loss of considerable volume and number of cells. Ideally, specimens need to be processed within 2 hours after collection. Keeping samples cold (at 4°C) for up to 24 hours has been shown to decrease the number of cells but not influence differential cell counts in humans and horses.15,28,30 Keeping BALF at ambient temperature and analyzing it within 4 hours does not affect cell counts nor lead to bacterial overgrowth.28 In the present study, BALF was centrifuged at 4°C within 15 minutes or was stored at 4°C and cytocentrifuged within 6 hours. An additional factor influencing cell lysis is lack of rigidity of the suction catheter that links the endoscope to the aspiration pump. If the catheter is too flexible, it collapses under aspiration and leads to sudden release of fluid and cells, potentially inducing cellular damage.
Variations in the speed and duration of centrifugation influence differential cell count. Lower cytocentrifugation speed has been shown to cause greater lymphocyte loss than higher speeds, while macrophage counts are affected in the opposite way.9,11,26 Indeed, De Brauwer and coworkers11 determined that the mean lymphocyte count obtained was significantly higher at 1200 rpm than at 500 rpm, while the number of both cell types was stable at speeds between 1200 and 2000 rpm. Consequently, a short period of high-speed cytocentrifugation is recommended to minimize lymphocyte loss.27 The speed of cytocentrifugation used in the present study was 1400 rpm, which corresponds well with the speeds reported in the literature. While specific recommendations concerning the optimum centrifugation speeds are lacking for most standard laboratory centrifuges, suggestions are that these should be the same as those that are best for cytocentrigutation.11 The acceleration rate during centrifugation does not appear to influence the differential cell count.9,11 Loss of lymphocytes may also occur during staining. This effect has been shown to be more of a problem with aqueous-based stains than with alcohol-based stains.10
Counting 300 cells in cytocentrifuged BALF samples, preferably in a circular pattern around the center of the cytocentrifuged spot, has been shown to lead to reliable enumeration of neutrophils, macrophages, lymphocytes, and eosinophils.12,16 In the present study, it was sometimes difficult to find 100 cells on pellet smears from individual healthy cats; that is why a differential cell count was established based on the examination of 100 cells on all preparations. Exclusion of isolated nuclei is particularly important, even when poorly cellular slides are being assessed, because isolated nuclei can be misinterpreted as lymphocytes. For this reason, in addition to the difficulty in counting 100 well-preserved cells, it was sometimes impossible to evaluate MSPs from animals with poor cellular content. In such cases, evaluation of CPs rather than MSPs is strongly advised, since this will give more reliable differential cell counts.
Indeed, not surprisingly, CPs in this study were easier to interpret. Previous studies in horses have shown that cellular morphology is consistently excellent on CP, while cells on MSP have poorer morphology, making identification (particularly of the small cells) more difficult.13,20 Cytocentrifugation also causes a uniform distribution of cells on a slide, whereas MSP produces a much more uneven cellular distribution pattern; the ease of counting is generally greater at the leading edge of the smear.20
Although previous studies failed to identify any significant difference in neutrophil counts between CP and MSP, the present study showed significantly higher percentages of neutrophils with CP than with MSP.20 In particular, dogs with bacterial respiratory disease had significantly higher neutrophil counts on CP slides than on MSP slides.
Previous studies of inter- and intraobserver repeatability demonstrated a good repeatability for all cell types except lymphocytes.32 In the present study, no significant observer effect was found for any type of cells.
Nine healthy cats were in the present study, and the mean total cell count obtained in samples from these animals was higher than values (which range from 270 to 301 cells/μL) reported in the literature.33–35 However, inconsistent BAL methodology was used in these studies. In the samples from healthy cats, macrophages were the predominant cell type, followed by neutrophils or eosinophils.33–37
Total cell count was increased in both nonbacterial inflammatory and bacterial disorders in both species, compared to values obtained in healthy cats [Table 2] or described in healthy dogs in the literature.1,8,31,38–43 Most differences were significant despite high variability in diseased animals. In dogs, significantly higher total cell count values were found in BALF from animals with bacterial disease of the lower respiratory tract. In the present study, total cell counts in dogs and cats with lower respiratory tract diseases were >866 cells/μL and >950 cells/μL, respectively. Total cell count values >1500 cells/μL have been reported in dogs with spontaneously arising lung disease, and values >595 cells/μL, up to 1500 cells/μL, in diseased cats have been reported.1,5 The significant difference in total cell counts between these groups is of minimal relevance, however, since it is probably a consequence of the criteria used to assign animals to each group.
In the present study, cats with bacterial respiratory disease had a significantly greater proportion of neutrophils in BALF than healthy cats; this finding is in agreement with published literature.7 Cats with nonbacterial inflammatory respiratory disease have an increase in either or both neutrophils and eosinophils.7,44,45 In a recent retrospective study of feline bronchial disease, eosinophilic inflammation occurred in 31% of the cases, and the disease accounted for 73% of eosinophilic BALF samples.7,45
Strict classification and universally accepted guidelines for interpretation of BALF are lacking. Such samples should only be interpreted in light of other clinical findings. A poor correlation often exists between the pathologist’s prediction of the types of cells likely to be present in the BALF and the cells that actually are present in the BALF based on histopathology results.4,5 Although the presence of a high number of lymphocytes possibly might suggest an immune reaction, and a high percentage of eosinophils might indicate a hypersensitivity reaction, in many cases the interpretation of the differential cell count is less obvious.5,14,39 As an example, a high number of lymphocytes in BALF can also be associated with lymphosarcoma.46
According to Hawkins and coworkers,3 neutrophilic, eosinophilic, or lymphocytic inflammation can be diagnosed by relative cell counts ≥12%, 14%, or 16% of these cell types within the total number of cells in a fluid, respectively. For Clercx and coworkers,21 eosinophilic bronchopneumopathy was best characterized by the presence of >50% eosinophils in BALF, although the disease can be diagnosed with significantly lower eosinophil counts.
In the present study, clinical diagnoses were classified into two groups: nonbacterial inflammatory and bacterial diseases. Diagnoses were based on clinical, radiographical, hematological, bronchoscopical, microbiological, and cytological findings. The criteria used to select animals with bacterial disease were relatively stringent since they included the presence of an alveolar pattern on thoracic radiographs and significant bacterial growth from BALF. These criteria are not a prerequisite for a diagnosis of bacterial bronchopneumonia, since they are not encountered in all cases. In agreement with the published literature, bacterial respiratory diseases in the present study were characterized by a significant increase in neutrophils in the BALF, while nonbacterial inflammatory respiratory diseases were characterized by a majority of macrophages and a less-marked increase of neutrophils.
Red blood cells were occasionally present in BALF samples that appeared macroscopically contaminated with blood. In such cases, at least part of the white blood cells counted and described in the slides from the BALF may have been from this blood contamination. Epithelial cells were more numerous in BALF from dogs with respiratory disease compared to those in BALF from diseased and healthy cats, and cells appeared more damaged on MSP compared to CP. Although the presence of ciliated cells may indicate that areas sampled are large airways rather than deeper regions, this could also be related to increased epithelial exfoliation in the aforementioned cases.14
Conclusion
The present study compared the cytological evaluation of BALF processed with CP versus MSP. Both methods, performed in appropriate standardized conditions, can provide specimens of good quality, allowing the clinician to obtain valuable diagnostic information. Since better quality was achieved with CP in all cases, showing a more homogeneous cellular distribution, this method should be used preferentially when possible. Cytospin preparation is particularly advised in the evaluation of BALF with a low total cell count. Since the type of smear significantly influences neutrophil differential count, the same technique should be consistently used in the monitoring of disease progression as well as in the assessment of the response to therapy.
Shandon Cytospin 4; Thermo Electron Corporation, Laborimpex, 1190, Brussels, Belgium
Citation: Journal of the American Animal Hospital Association 44, 6; 10.5326/0440285
Citation: Journal of the American Animal Hospital Association 44, 6; 10.5326/0440285
Citation: Journal of the American Animal Hospital Association 44, 6; 10.5326/0440285
Citation: Journal of the American Animal Hospital Association 44, 6; 10.5326/0440285
Citation: Journal of the American Animal Hospital Association 44, 6; 10.5326/0440285
Activated alveolar macrophage. This is a larger cell, with more abundant cytoplasm and more extensive cytoplasmic vacuolation. To the left of this cell is a binucleate macrophage, a further feature of activation. Strands of eosinophilic mucus are seen in the background. This cytospin sample is from a dog with nonbacterial inflammatory respiratory disease.
Bronchoalveolar lavage fluid (BALF) neutrophils. This is a cytospin preparation of mature, segmented neutrophils within BALF from a dog with bacterial respiratory disease.
Bronchoalveolar lavage fluid (BALF) eosinophils. Eosinophils (arrowed) are the predominant cell type in this cytospin sample of BALF from a cat with nonbacterial inflammatory respiratory disease. The sample includes fewer segmented neutrophils and a cell with plasmacytoid morphology (arrowhead).
Unstimulated and reactive lymphocytes in bronchoalveolar lavage fluid from a dog with nonbacterial inflammatory respiratory disease. The small lymphocytes (arrow) have round nuclei and scant surrounding cytoplasm. The larger lymphocyte (arrowhead) has more abundant cytoplasm and a more oval, slightly indented nucleus. The sample is a cytospin preparation.
Ciliated respiratory epithelial cells in bronchoalveolar lavage fluid from a dog with nonbacterial inflammatory respiratory disease. This field shows several individual columnar epithelial cells with basal nuclei and surface cilia. This is a smear from a cell pellet.
