Abstract
Empiric antibiotic therapy is often employed to treat feline chronic rhinosinusitis (CRS), however, collection of samples for culture should result in improved antibiotic selection and development of less antibacterial resistance. This study evaluated the effect of sampling method on culture results in 44 cats and found that aerobic and anaerobic cultures were positive significantly more often from nasal flush samples than from tissue biopsy samples. Frequency of positive Mycoplasma species culture did not differ between the two sampling methodologies, although results were discordant in eight sample pairs. Mycoplasma species were cultured from flush samples only in three cats and from biopsy samples only in five cats. These results suggest that a nasal flush sample is adequate for determining the presence of bacterial organisms in feline nasal disease, however, failure to culture Mycoplasma species from a nasal flush must be interpreted with caution.
Feline chronic rhinosinusitis (CRS) is a disease of unknown etiology associated with high morbidity in cat populations. Purported causes include viral infection or virally induced damage to nasal turbinates, local activation of inflammation due to various micro-organisms, and/or dysregulated mucus production. Several studies have indicated that most cases of upper respiratory tract disease in the cat are complicated by secondary bacterial infection, 1–5 which explains the partial antibiotic responsiveness of this disorder. A prospective study evaluated the role of bacterial and viral organisms in affected cats and described greater isolation rates of potentially pathogenic aerobic bacteria from cats with CRS in comparison to control cats. 3 In that study, results were based on microbiologic culture of a nasal flush sample, although nasal swab samples have been used in other studies. 4–6
In a case-control study of microbial isolation comparing cats with CRS to controls, overall detection rates for aerobes, anaerobes, and Mycoplasma species did not significantly differ between nasal flush and biopsy samples. 3 This might suggest that a nasal flush sample is an appropriate sampling method for detection of clinically significant organisms. However, only 17 cats were included in that study, and anaerobic bacteria and Mycoplasma species were isolated infrequently. Therefore, the objective of the present study was to compare the results of microbial culture of samples collected by nasal flush or biopsy in a large number of cats with clinical signs of upper respiratory tract disease.
Materials and methods
Cats entered into the study had mucopurulent nasal discharge for >7 days and were deemed by one of three local animal shelters to be unadoptable due to respiratory disease. Antibiotic history was not available for the majority of cats and, therefore, was not considered for this study. Immediately after euthanasia with intravenous barbiturate overdose, nasal flush and biopsy samples were obtained from the left side of the nasal cavity. To obtain a surface sample for microbial assessment, nasal flush was performed using a sterile 8-French red rubber catheter passed from the naris into the caudal aspect of the nasal cavity. The caudal nasopharynx was obstructed by digital compression, 3 ml of sterile saline were injected through the catheter, and an aliquot was retrieved through aspiration. The sample was immediately placed in semisolid anaerobic culture medium (Anaerobe Systems, Morgan Hills, CA) for aerobic, anaerobic, and Mycoplasma species cultures. A pinch biopsy was obtained following the nasal flush from the same side of the nasal cavity using a 3 mm cup biopsy instrument (Karl Storz Veterinary Endoscopy, Goleta, Cal). The biopsy was obtained blindly from the caudal aspect of the nasal cavity and immediately placed in semisolid anaerobic culture medium (Anaerobe Systems, Morgan Hills, CA) for aerobic, anaerobic, and Mycoplasma species cultures. All procedures were approved by the Institutional Animal Care and Use Committee at the University of California, Davis (UCD).
Nasal flush and biopsy samples were processed at the Veterinary Microbiology Laboratory of the UCD Veterinary Medical Teaching Hospital (VMTH) within 2 h of collection. Flush samples were plated onto 5% sheep blood agar and MacConkey's agar for isolation of aerobic organisms, pre-reduced anaerobic Brucella species plates (Anaerobe Systems, Morgan Hill, CA) for anaerobic culture, and pleuropneumonia-like organism base with thallium acetate (antifungal) and penicillin G (antibacterial) for isolation of Mycoplasma species (UC Davis Media Lab). Biopsy samples were placed in 1 ml of brain heart infusion and macerated using sterile technique prior to culture for aerobes, anaerobes, and Mycoplasma species using the same media described for flush samples. Bacterial growth was assessed in a semi-quantitative fashion by counting the number of quadrants with bacterial growth and was reported as 1+, 2+, 3+, or 4+. Standard biochemical methods were used to identify cultured bacteria. 7 Gram-negative bacilli that produced oxidase, did not ferment glucose, sucrose, or lactose, and did not produce indole, gas, or H2S (ie, triple sugar iron reaction 5) but which were not classified as Bordetella bronchiseptica, Pseudomonas species, or Acinetobacter species were described as non-enteric Gram-negative rods. Plates for Mycoplasma species culture were incubated at 37°C in 5% CO2 for 5 days and identified as probable Mycoplasma species if a Gram stain revealed poorly stained Gram-negative pleomorphic organisms and colony morphology examined under a dissecting microscope was of a classic ‘fried egg’ appearance. Plates used for isolation of anaerobic bacteria were inspected daily for 1 week to detect growth of bacteria.
Statistical analysis
Concordance and discordance of culture results were compared between pairs of flush and biopsy samples by use of McNemar's χ2 test. Results for aerobic bacteria, anaerobic bacteria, and Mycoplasma species were examined separately. For all analyses, P≤0.05 was considered significant. Using results of the nasal flush as the gold standard for organism detection, biopsy sensitivity (proportion of positive nasal flushes found positive on biopsy), biopsy specificity (proportion of negative nasal flushes found negative on biopsy), and biopsy accuracy (proportion of positive and negative nasal flushes found positive and negative, respectively, on biopsy) were calculated. Sensitivity, specificity, and accuracy are reported with 95% confidence intervals (CI).
Results
Paired nasal flush and biopsy samples were collected from 44 cats that met inclusion criteria. These samples were submitted for aerobic and Mycoplasma species cultures for all 44 cats; but anaerobic cultures were performed for only 39 cats due to laboratory error. Isolation rates for all organisms using the two sampling methods are summarized in Table 1.
Summary of isolation rates for aerobes, anaerobes, and Mycoplasma species in nasal flush and biopsy samples from cats
Considering both sample methods together, aerobic bacteria were cultured from a total of 54/88 (61%) samples. In 27 samples, one species of aerobic bacteria was cultured, two species were cultured in 20 samples, and three species were cultured in seven samples. The aerobic bacteria most frequently isolated were non-enterics (n=20), Staphylococcus species (n=18), Pasteurella species (n=10), E coli (n=6), and Streptococcus species (n=3). Aerobic bacteria were isolated significantly more often (P=0.001) from nasal flush samples (33/44; 75%) than from biopsy samples (21/44; 48%). Overt qualitative differences in bacterial species isolated were not noted between samples collected by flush and those collected by biopsy. All cats that had three species of aerobic bacteria isolated from a flush sample (n=5) had two or all three of the same species isolated from the paired biopsy sample. One cat in which the nasal flush sample contained Corynebacterium ulcerans, had both C ulcerans and Staphylococcus species cultured from the biopsy sample; however, in the remainder of cats, the same organisms were isolated from flush and tissue samples. Considering only whether an aerobic organism was cultured or not, results for paired biopsy and flush samples were concordant for 30/44 cats; of these, 10/44 cats had no aerobic bacteria isolated from either sample. In 13/14 discordant results, flush samples produced aerobic bacterial growth when biopsy samples did not. Growth of aerobic bacteria was reported as 1+ in 19 flush and 14 biopsy samples, 2+ in nine flush and six biopsy samples, 3+ in four flush and zero biopsy samples, and 4+ in one flush and biopsy sample.
Anaerobic bacteria were cultured from only 6/78 (8%) samples, all but one of which were flush samples. Thus, anaerobic bacteria were cultured significantly (P=0.05) more often from nasal flush samples (5/39, 13%) than from biopsy samples (1/39, 3%). Considering only whether an anaerobic organism was cultured or not, results for paired biopsy and flush samples were concordant in 35/39 cats. In all four cats with discordant results, flush samples produced anaerobic bacterial growth when biopsy samples did not. A limited number of anaerobic species were isolated. Bacteroides/Prevotella species were isolated from all flush samples. Peptostreptococcus and Fusobacterium species were isolated in conjunction with Bacteroides species in two other flush samples. Bacteroides ureolyticus was cultured from the only flush and biopsy sample pair for which culture results were positive and concordant.
Mycoplasma species were cultured from a total of 30/88 (34%) samples; approximately equal numbers of flush (n=14) and biopsy (n=16) samples yielded positive growth.
Mycoplasma species culture results for paired biopsy and flush samples were concordant in 36/44 cats. Of the discordant results, 3/14 (21%) flush samples yielded Mycoplasma species when biopsy samples did not, and 5/16 (31%) biopsy samples yielded Mycoplasma species when flush samples did not. There was no significant difference in the rate of detection of Mycoplasma species between flush and biopsy samples (P=0.54).
When culture results from nasal flush were used as the gold standard, specificity was high for all culture results from biopsy samples, and accuracy of biopsy sample results was high for both Mycoplasma species and anaerobic cultures. Sensitivity was highest for Mycoplasma species culture results from biopsy tissue and lowest for anaerobic bacterial culture results from biopsy samples (Table 2).
Sensitivity, specificity, and accuracy of biopsy in detecting aerobic and anaerobic bacterial and Mycoplasma species using results from culturing nasal flushes as the gold standard. 95% CI are listed in parentheses
Discussion
In this study, cultures of nasal flush samples for aerobic and anaerobic bacteria were positive significantly more often than cultures of tissue biopsy samples, and in cases where both samples yielded positive growth, the organisms found in the specimens were frequently identical. This suggests that nasal flush is the preferred technique for determining bacterial presence in feline nasal disease and maximizes detection of potentially pathogenic bacterial organisms. However, it does not necessarily confirm that nasal flush is the preferred technique, as the diagnostic utility of each test was not directly assessed in this study. Other clinically relevant factors should be considered when choosing between these sampling techniques. For example, obtaining a tissue biopsy sample requires general anesthesia and may cause prolonged epistaxis, while the collection of a nasal flush can be performed under heavy sedation and is less traumatic to the nasal mucosa. However, feline nasal disease is frequently characterized by deep tissue inflammation and turbinate destruction, 3 which requires biopsy results to diagnose and interpret. Therefore, even though results of the present study suggest that assessing a nasal biopsy sample for evidence of infection is not required to obtain adequate aerobic or anaerobic bacterial culture results, the use of flush and biopsy may provide complementary histologic and bacteriologic data when assessing patients with CRS.
In contrast to aerobic and anaerobic bacterial detection rates, culture results for Mycoplasma species did not differ between flush and biopsy samples. It is of note that results were discordant in 8/44 sample pairs, with culture of biopsy samples more frequently yielding Mycoplasma species than did flush samples. While not statistically significant, the discordant results obtained with different sampling methods could be clinically relevant because of the need to identify this organism and choose an appropriate antibiotic for this cell-wall deficient bacterium. Failure to culture Mycoplasma species from a nasal flush might not be a true reflection of the absence of Mycoplasma species in the nasal cavity because a tissue sample may be required to isolate the organism. One reason for this might be that Mycoplasma species can be found in very close apposition to epithelial cells, with some species exhibiting specialized organelles to promote adherence to cells. 8 Some mycoplasmas can invade and infect epithelial or mesenchymal cells that are not naturally phagocytic.9,10 Thus, it is possible that a deep nasal flush might not result in sufficient exfoliation of organisms or cells to result in a positive culture result. Discordant findings in nasal flush and biopsy samples may also be clinically relevant because of evidence that a negative Mycoplasma species culture does not always reflect true absence of this species in the sample since not all Mycoplasma species are cultivable. A previous study revealed that M arginini could be detected in a nasal flush sample by use of the polymerase chain reaction (PCR) despite that sample being negative on culture. 11 Therefore, it is possible that culture results of flush and biopsy pairs from the present study were discordant because different Mycoplasma species were present in the two samples.
In a previous study comparing PCR to culture for Mycoplasma species detection in nasal samples of cats, 11 concordant results were reported for the two techniques. Considering those results in conjunction with data from the present study, it is possible that while culture of a nasal flush is appropriate for detection of aerobic and anaerobic bacteria, PCR assessment of a flush and/or biopsy sample is required for more accurate assessment of Mycoplasma species status. Finally, some Mycoplasma species such as M gateae and M arginini are considered saprophytes or normal flora of mucosal surfaces 12 while others, particularly M felis, are considered pathogenic because increased isolation rates are reported in diseased cats compared to clinically healthy cats.13,14 This suggests that quantitative PCR followed by nucleotide sequencing for speciation would be the optimal method to assess the clinical relevance of Mycoplasma species found in nasal samples.
Previous cytologic studies in feline nasal disease revealed that brush sampling techniques provide an adequate representation of the overlying mucosal layers, but are less accurate in detecting inflammatory or neoplastic changes in the deeper mucosal layers.6,15 The same is likely true in comparing culture results of flush and biopsy samples in the current study. The microbial data obtained here could indicate that cultures of nasal flush samples reflect only surface bacterial organisms and that deeper tissues are not always infected with these organisms. However, it is also possible that those cultures of tissue biopsy samples that displayed no growth did so because of the manipulation required to plate the samples. While flush samples could be applied directly to appropriate media, tissue samples required maceration prior to plating, which may have resulted in loss of bacterial viability. Also, the additional tissue present in biopsy samples may have lead to dilution of bacterial numbers or may have contained endogenous or exogenous inhibitors of bacteria, resulting in inadequate viable organisms for culture.
One limitation of the current study is the lack of information regarding antibiotic administration in cats examined here. Depending on the spectrum of the drug used and duration of treatment, it is possible that negative culture results obtained here were the result of previous antibiotic treatment. However, as samples obtained by different methodology most often were positive for the same aerobic or anaerobic bacterial organism, it is unlikely that the lack of data impacts the comparison between methodologies described here.
Results presented here suggest that interpretation of the role of bacteria and particularly of Mycoplasma species in feline CRS must consider the method used for sample collection. Although conventional culture techniques on a non-invasively obtained nasal flush sample can be used to identify the presence of aerobic or anaerobic bacterial organism, molecular techniques may be required to determine whether mycoplasmal species are involved. Due to discordant results obtained with paired nasal flush and biopsy samples in the study performed here, further research is required to determine the appropriate sampling method to assess the role of Mycoplasma species in feline nasal disease.
Footnotes
Acknowledgements
Funding for laboratory testing was supplied by grants from the Center for Companion Animal Health University of California, Davis (01-05-F) and the San Francisco Foundation, University of California, Davis (02-69-F). The authors would like to thank Dr David J Maggs for review of this manuscript.