Although BRD has been extensively studied for many years, it remains the most costly disease of the beef cattle industry,1 and knowledge gaps regarding its epidemiology still exist. Bovine respiratory disease has a complex causal web with multiple contributing factors including host immunity, stressors, and viral (bovine respiratory syncytial virus, bovine herpes virus type 1, parainfluenza virus type 3, and bovine viral diarrhea virus) and bacterial (Mannheimia haemolytica, Pasteurella multocida, and Histophilus somni) pathogens.2 Those 3 bacterial species are commonly implicated, either together or separately, in BRD outbreaks, and diagnostic testing modalities to identify those bacteria and determine their in vitro antimicrobial susceptibility can be used to formulate treatment and control regimens for diseased cattle and cattle at high risk of developing disease (ie, high-risk cattle).
Various antemortem and postmortem methods for obtaining samples from the respiratory tract of cattle for bacterial culture have been investigated and include nasal swabs, transtracheal swabs, transtracheal washes, NPSs, BALs, lung lavages, lung swabs, and lung tissue specimens.3–9 Each method has strengths and limitations and varies in terms of the materials and equipment necessary for sample collection, extent and type of animal restraint required, degree of invasiveness, time and skill needed to obtain samples, and the area of the respiratory tract from which samples are obtained. For example, in live cattle, collection of samples from the URT is easier than collection of samples from the LRT because it requires fewer materials and less time and skill. However, if a sample from the LRT of a live animal is desired, a blind (ie, performed without endoscopic guidance) BAL is quicker and more practical and economical to perform than sample collection methods that require an endoscope or transtracheal wash, especially if samples need to be collected from multiple calves without cross-contamination during a short period of time. Additionally, BAL is less invasive than a transtracheal wash, and cattle generally tolerate BAL well without sedation. An advantage of performing BAL with endoscopic guidance is that samples can be obtained from a specific lung lobe. In calves with BRD, the right cranial lung lobe is commonly affected5 and samples obtained from that lobe may yield different diagnostic test results than samples obtained from less severely affected lobes. An important disadvantage of collecting BALF samples with endoscopic guidance is that the endoscope must be disinfected between animals, whereas only 1 sterile tube is necessary for collection of a blind BALF sample. A disadvantage of collecting samples by both blind and endoscopic-guided BAL is the potential for introducing microorganisms from the nasal passages into the lungs. That risk can be partially mitigated by use of a guarded endoscope; however, to our knowledge, a comparison between guarded and unguarded methods for BAL and determination of the impact each method has on the bacterial flora of the LRT have not been performed in cattle. Regardless of the method used or the location from which a sample is obtained, all samples are subject to the limitations associated with culturing microorganisms and the potentially complex ecology of the normal microbial flora of the respiratory tract of individual cattle.10
The purpose of the present study was to assess the use of bilateral NPSs and BAL for obtaining samples from the URT and LRT, respectively, of live cattle. Both of those sample collection methods can be done easily and quickly in a field setting with basic restraint equipment such as a squeeze chute and halters, are much less invasive than other methods, are fairly well tolerated by calves, and require minimal technical expertise. However, the scientific literature contains conflicting results regarding the extent of agreement between bacterial culture results of samples obtained from the URT and LRT and how accurately those results reflect the organisms present in the respiratory tract.4–6 To our knowledge, bacterial culture results obtained from pooled bilateral NPS and BALF samples have not been compared. In the study reported here, we compared bacterial culture results, gamithromycin susceptibility profiles, PPVs and NPVs, and agreement between pooled NPS and BALF samples for identification of various M haemolytica genotypes, P multocida, and H somni in beef calves that were treated for BRD.
Materials and Methods
Animals
The calves evaluated in the study reported here were enrolled in a larger study,3 which consisted of 180 commingled male mixed-breed beef calves that were considered at high risk of developing BRD. All samples obtained from the cattle evaluated in this study were collected as part of the larger study,3 which was approved by the Kansas State University Institutional Animal Care and Use Committee. The 28 calves assessed in this study developed BRD and were treated with the labeled dose of gamithromycin (6 mg/kg, SC, once; the dose was divided as necessary so that no more than 10 mL of the antimicrobial was administered per injection site). Prior to gamithromycin administration, each calf was randomly assigned to 1 of 3 sampling groups as described.3 For the calves of group 1 (n = 9), respiratory tract samples were obtained by NPS and BAL immediately before (day 0) and 5 days (day 5) after gamithromycin administration. For the calves of group 2 (n = 10), respiratory tract samples were obtained by NPS on days 0, 0.5, and 5 and by BAL on days 0.5 and 5. For the calves of group 3 (n = 9), respiratory tract samples were obtained by NPS on days 0, 1, and 5 and by BAL on days 1 and 5. Therefore, a total of 56 paired NPS and BALF samples were obtained. Data from 2 gamithromycintreated calves that did not finish the posttreatment interval and were excluded from the other study3 because they could not be classified as a treatment success or failure were included in this study because samples were obtained from both calves 5 days after gamithromycin administration.
NPS method
At each NPS sample collection, a sterile double-guarded uterine culture swaba was passed up each nostril and guided through the nasal passage until resistance was encountered at a point approximately equivalent to the distance between the medial canthus of the ipsilateral eye and the end of the muzzle. Then, the entire swab apparatus was retracted a short distance so that the culture swab could be advanced through both the outer and inner guards and rotated against the nasopharyngeal mucosa; after several rotations were made, the swab was retracted back through the inner and outer guards and removed from the nasal cavity. The swab portion was then placed into a vial of liquid Amies culture medium and stored on ice prior to bacterial culture. A separate swab was used for each nostril, and the swabs from both nostrils were placed in the same vial for transport to the laboratory; thus, the NPS samples obtained from each calf during a particular sample acquisition were considered pooled.
BAL method
Bronchoalveolar lavage fluid samples were obtained by use of a technique adapted from a previous description.8 Briefly, an unsedated calf was restrained in a squeeze chute and 2 rope halters were used to lift and extend the head so that the nasal bone was parallel to the ground. A 240-cm BAL catheterb was then inserted through a naris and blindly guided through the nasal passage into the trachea until the end was wedged in a bronchus. Correct placement of the catheter was verified by elicitation of the coughing reflex, movement of air into and out of the catheter with each breath, and the absence of rumen contents, odor, and gurgling from the catheter. Once wedged in the appropriate location, the cuff on the catheter was left uninflated and a total of 240 mL of sterile saline (0.9% NaCl) solution was infused into and immediately aspirated from the bronchus in 3 phases or doses; the first, second, and third doses consisted of 120, 60, and 60 mL of saline solution, respectively. Approximately 135 to 175 mL of BALF was obtained from each calf and was divided equally into 4 conical vials. Each vial was kept on ice until centrifugation within 40 minutes after collection; one of the resulting cell pellets was randomly selected and resuspended in liquid Amies medium for bacterial culture.
Bacterial culture and determination of gamithromycin susceptibility
The NPS and BALF samples were processed and underwent bacterial culture as described.3 A sample was considered positive for M haemolytica, P multocida, or H somni if the culture yielded at least 1 colony of the specified bacterium. The gamithromycin susceptibility for up to 12 M haemolytica and 6 P multocida colonies from each culture-positive sample was determined by a broth microdilution method as described,3 whereas the gamithromycin susceptibility for up to 6 H somni colonies from each culture-positive sample was determined by use of a microdilution method that was slightly modified from the guidelines established by the Clinical and Laboratory Standards Institute.11 Briefly, H somni isolates were cultivated on chocolate II agar plates at 37°C for 24 hours in an atmosphere that contained 5 ± 2% CO2. For each isolate, the resulting bacterial colonies were transferred in a quantity sufficient to achieve an optical density equivalent to 0.5 McFarland standard to a 5-mL tube containing cation-adjusted Mueller-Hinton broth. That suspension was used to inoculate 2× veterinary fastidious mediumc (15 μL/mL), which was then dispensed into the wells (50 μL/well) of a 96-well custom frozen susceptibility platec for gamithromycin. The plate was sealed with sealing film and incubated at 37°C for 24 hours in an atmosphere with 5% CO2. Following incubation, the plate was visually inspected, and the minimum inhibitory concentration for gamithromycin was determined as the lowest concentration of the drug that prevented visible bacterial growth. Current Clinical and Laboratory Standards Institute guidelines11 were used to classify the gamithromycin susceptibility of each isolate as susceptible, intermediate, or resistant.
Genomic sequencing, bioinformatics, and phylogenetic analysis of M haemolytica
An SNP-based typing method was developed from M haemolytica isolates characterized from the calves of this study and other sources as described elsewhere.12 Briefly, 1 colony of each M haemolytica isolate was suspended in 1 mL of brain-heart infusion broth and placed in a well of a 96-deep well plate and cultivated overnight (approx 16 to 20 hours) without shaking. Isolate DNA was extractedd and quantified by use of a fluorometere in accordance with the manufacturer's instructions. Deoxyribonucleic acid libraries were constructedf and sequenced.g A minimum of 10× sequence reads/genome was obtained for each isolate sequenced. Each library was mapped to a closed circular M haemolytica genome in Gen-Bank (CP004752)13 by use of a publicly available automated mapping algorithmh as described.12 Neighbor-joining phylogenetic trees were constructed from concatenated SNP genotypes of each isolate by use of a bootstrapping method and F84 substitution model.i The phylogenetic trees were generated and graphically displayed with publicly available softwarej as described.14
Data analysis
Bacterial culture results for all NPS and BALF samples obtained from each calf were summarized. Because the study population was small (n = 28 calves), the culture results for each of the 3 bacterial organisms evaluated (M haemolytica, P multocida, and H somni) for the paired NPS and BALF samples that were collected 5 days after gamithromycin administration were cross tabulated in 2 × 2 tables. The culture results for the BALF samples were considered the gold standard, and the PPVs and NPVs were calculated for the culture results of the NPS samples as described.15 Additionally, the kappa statistic (κ) was calculated as described15 for each of the organisms evaluated to compare the extent of agreement beyond that expected by chance between the culture results obtained for NPS and BALF samples. A κ ≤ 0 was indicative of poor agreement, from 0.01 to 0.2 was indicative of slight agreement, from > 0.2 to 0.4 was indicative of fair agreement, from > 0.4 to 0.6 was indicative of moderate agreement, from > 0.6 to 0.8 was indicative of substantial agreement, and from > 0.8 to 1.0 was indicative of almost perfect agreement.16 To evaluate the potential for bias between NPS and BAL, the exact McNemar significance probability15 was determined for each κ statistic, and values ≤ 0.05 were considered significant evidence of bias. Predictive values and κ values were not calculated for the other sample acquisition times because the number of calves from which paired NPS and BALF samples were collected was too small (≤ 10 calves).
Results
Bacterial culture results
A total of 75 NPS samples were collected from the 28 study calves and underwent bacterial culture, of which 22 (29%) were positive for M haemolytica, 35 (47%) were positive for P multocida, and 14 (19%) were positive for H somni. Fifty-six BALF samples were collected and underwent bacterial culture, of which 10 (18%) were positive for M haemolytica, 17 (30%) were positive for P multocida, and 6 (11%) were positive for H somni. The M haemolytica, P multocida, and H somni culture results for all NPS and BALF samples obtained from each calf were tabulated (Supplemental Tables S1 and S2, available at: http://avmajournals.avma.org/doi/suppl/10.2460/ajvr.78.3.350). Culture results varied over time within individual calves, sometimes in an unexpected manner. For example, several calves were culture negative for a specific pathogen prior to gamithromycin administration and became culture positive for that pathogen following antimicrobial treatment.
Agreement and predictive values for bacterial culture results of paired NPS and BALF samples
Of the 56 paired NPS and BALF samples that were collected during all 4 sample acquisition times, both specimens were culture positive for M haemolytica, P multocida, or H somni in 10, 15, and 4 pairs, respectively. Mannheimia haemolytica, P multocida, and H somni culture results for paired NPS and BALF samples collected 5 days after gamithromycin administration were cross tabulated (Table 1). The PPV and NPV of NPS samples for M haemolytica were 67% and 100%, respectively, and there was substantial agreement (κ, 0.73) between the M haemolytica culture results for NPS and BALF samples. The exact McNemar significance probability for M haemolytica was 0.25. The PPV and NPV of NPS samples for P multocida were 75% and 100%, respectively, and there was almost perfect agreement (κ, 0.81) between the P multocida culture results for NPS and BALF samples. The exact McNemar significance probability for P multocida was 0.50. The PPV and NPV of NPS samples for H somni were 100% and 96%, respectively, and there was substantial agreement (κ, 0.78) between H somni culture results for NPS and BALF samples. The exact McNemar significance probability for H somni was 1.0.
Cross tabulation of Mannheimia haemolytica, Pasteurella multocida, and Histophilus somni culture results for NPS and BALF samples obtained from 28 male mixed-breed beef calves 5 days after they were administered gamithromycin (6 mg/kg, SC, once) for the treatment of BRD.
| NPS sample culture result | |||
|---|---|---|---|
| Bacterial organism | BALF sample culture result | Positive | Negative |
| M haemolytica | Positive | 6 | 0 |
| Negative | 3 | 19 | |
| P multocida | Positive | 6 | 0 |
| Negative | 2 | 20 | |
| H somni | Positive | 2 | 1 |
| Negative | 0 | 25 | |
Values represent the number of calves with the given culture result.
Whole-genome sequencing of M haemolytica
Two hundred eighty-seven separate colonies of M haemolytica were retained for whole-genome sequencing from 22 and 10 culture-positive NPS and BALF samples, respectively, that were obtained during all 4 sample acquisition times. Genetic subtype information was not discernable for 11 isolates. The remaining 276 M haemolytica isolates were split into 2 clearly discernable phylogenetic clades, each of which was further divided into subtypes (Figure 1). For each of the 10 paired NPS and BALF samples in which both specimens were culture positive for M haemolytica, the same subtype was isolated from both specimens (Supplemental Table S3, available at http://avmajournals.avma.org/doi/suppl/10.2460/ajvr.78.3.350). None of the culture-positive BALF samples and only 1 culture-positive NPS sample contained > 1 subtype of M haemolytica. The NPS sample from which 2 subtypes of M haemolytica were isolated was obtained from a calf immediately prior to gamithromycin administration.
Neighbor-joining phylogenetic trees for Mannheimia haemolytica clades 1 and 2 as determined by genome-wide concatenated SNPs from 276 isolates subjected to whole-genome sequencing. All M haemolytica isolates were acquired from NPS or BALF samples that were obtained from calves with clinical BRD immediately before or 0.5, 1, or 5 days after treatment with gamithromycin (6 mg/kg, SC, once). Genetic subtypes within each clade are denoted by lower case letters. Numbers beside internal nodes of each tree represent bootstrap percentage values from 100 pseudo-alignments. The scale bar represents substitutions per site within each tree. The genetic distance between clades 1 and 2 exceeded the genetic distance within the 2 clades and is not shown.
Citation: American Journal of Veterinary Research 78, 3; 10.2460/ajvr.78.3.350
Gamithromycin susceptibility profiles
The gamithromycin susceptibility phenotype (susceptible, intermediate, or resistant) was in agreement for 7 of 10 paired NPS and BALF samples in which both samples cultured positive for M haemolytica, 13 of 15 paired NPS and BALF samples in which both samples cultured positive for P multocida, and all 4 paired NPS and BALF samples in which both samples cultured positive for H somni (Table 2). Within individual samples, the gamithromycin susceptibility phenotype varied for 4 of 22 NPS samples and 1 of 10 BALF samples that were culture positive for M haemolytica, 3 of 35 NPS samples and 2 of 17 BALF samples that were culture positive for P multocida, and 2 of 14 NPS samples and 0 of 6 BALF samples that were culture positive for H somni (Supplemental Tables S1 and S2).
Descriptive data regarding the extent of agreement in gamithromycin susceptibility results for paired NPS and BALF samples obtained over several days from the calves of Table 1, in which both samples were culture positive for M haemolytica, P multocida, or H somni.
| Gamithromycin susceptibility of paired samples | ||||
|---|---|---|---|---|
| Bacterial organism | Sample acquisition time (d) | No. of culture-positive paired NPS and BAL samples | Agreed | Disagreed |
| M haemolytica | 0 | 2 | 2 | 0 |
| 0.5 | 2 | 1 | 1 | |
| 1 | 0 | 0 | 0 | |
| 5 | 6 | 4 | 2 | |
| P multocida | 0 | 3 | 2 | 1 |
| 0.5 | 5 | 5 | 0 | |
| 1 | 1 | 1 | 0 | |
| 5 | 6 | 5 | 1 | |
| H somni | 0 | 1 | 1 | 0 |
| 0.5 | 0 | 0 | 0 | |
| 1 | 1 | 1 | 0 | |
| 5 | 2 | 2 | 0 | |
Samples acquired on day 0 were obtained immediately before gamithromycin administration. For each isolate, the MIC for gamithromycin was determined by use of a broth microdilution method, and Clinical and Laboratory Standards Institute guidelines were used to classify the susceptibility phenotype of each isolate as susceptible, intermediate, or resistant. Gamithromycin susceptibility agreed when the phenotype was the same for both paired samples and disagreed when the phenotype differed between the paired samples.
See Table 1 for remainder of key.
Discussion
In the present study, there was substantial or almost perfect agreement between bacterial culture results for M haemolytica, P multocida, and H somni obtained from NPS and BALF samples collected simultaneously from calves with BRD 5 days after they were treated with gamithromycin. The extent of agreement was measured by use of a κ, which is a useful measure when a gold standard test for the condition being evaluated is either not available (ie, has not been established) or feasible.15 The gold standard for diagnosis of BRD in cattle includes examination of lung pathology and culture of lung tissue for various pathogens during necropsy. However, a reliable method for BRD diagnosis in live cattle is necessary. According to data from Table 1 in 1 study,6 there was perfect agreement between the M haemolytica culture results for NPS and lung lavage fluid samples obtained from dead calves, whereas there was only substantial, but imperfect, agreement (κ, 0.73) between the M haemolytica culture results for NPS and BALF samples obtained from the live calves of the present study. It should be noted that a postmortem lung lavage is not equivalent to an antemortem BAL because a larger portion of the lungs can be lavaged, or sampled, after death than is feasible prior to death. In another study,5 there was imperfect agreement between bacterial culture results for unilateral NPS samples and BALF samples obtained with endoscopic guidance from feedlot calves with and without clinical signs of BRD prior to antimicrobial treatment. The investigators of that study5 concluded that, for individual calves, NPS samples did not accurately predict the bacteria isolated from concurrently collected BALF samples. In the present study, NPS samples obtained via both the right and left nasal passages were placed in the same vial with culture medium for transport to the laboratory. Thus, the 2 NPS samples obtained from each calf at each specimen collection time were essentially pooled, which may have enhanced our ability to isolate organisms from the URT, resulting in the substantial or almost perfect agreement in culture results between NPS and BALF samples. Also, although the κs for the extent of agreement between culture results of NPS and BALF samples for this study were higher than those for that other study,5 it is important to remember that the κ is partly influenced by prevalence,17 which may vary with disease or treatment status. Therefore, the κ for assessing the extent of agreement between 2 diagnostic tests may vary depending on whether clinical versus subclinically affected or treated versus untreated animals are evaluated. Ultimately, careful consideration of the sample population is important for interpretation of κs, and the extent of agreement between culture results for paired NPS and BALF samples is dynamic and will not be identical for all populations of cattle.
Of the paired NPS and BALF samples collected 5 days after gamithromycin administration, only 1 NPS sample was culture negative for H somni while the concurrent BALF sample was culture positive for H somni, resulting in an NPV of 96% for NPS samples when H somni culture results for BALF samples were considered the gold standard. The NPV of NPS samples for both M haemolytica and P multocida was 100%. This indicated that, if M haemolytica, P multocida, or H somni was not cultured from an NPS sample 5 days after treatment with gamithromycin, it was unlikely that pathogen would be cultured from a concurrently collected BALF sample. Interestingly, in several calves, M haemolytica or P multocida was cultured from the NPS sample but not the BALF sample, which resulted in PPVs < 100% when culture results for BALF samples were considered the gold standard. Those results conflict with data from another study,6 in which the PPV and NPV were both 100% when M haemolytica culture results for unilateral NPS samples were compared with those for postmortem lung lavage fluid samples (gold standard). Unfortunately, very few samples in that study6 were culture positive for P multocida and none were culture positive for H somni; therefore, the PPV and NPV of NPS samples for those 2 organisms could not be calculated in that population. Among the paired NPS and BALF samples collected on day 5 (5 days after gamithromycin administration) in the present study, 3 of 9 NPS samples that were culture positive for M haemolytica had paired BALF samples that were negative for that organism; likewise, 2 of 8 NPS samples that were culture positive for P multocida had paired BALF samples that were negative for that organism. Because those samples were obtained 5 days after gamithromycin administration, it is possible that the organisms were newly acquired from other calves and were present in the nasopharynx of the sampled calves but had not proliferated in the LRT in sufficient numbers for detection in the BALF samples. Alternatively, the organisms may have been present in a portion of the LRT that was not sampled by the BAL, or gamithromycin administration affected microbial growth in the URT and LRT differently. Also, given that M haemolytica, P multocida, and H somni have been proposed as commensals of the URT,18 it is possible that the isolation of those organisms from NPS samples but not BALF samples collected 5 days after gamithromycin treatment in some calves of the present study might simply reflect a return of the normal URT microflora in those calves. Regardless, the results of the present study indicated that culture of an organism from an NPS sample after treatment does not reliably predict that the organism will be cultured from a concurrently collected BALF sample (ie, the presence of an organism in the URT does not guarantee that it will also be present in the LRT). However, predictive values are dependent on prevalence.15 Thus, for any given diagnostic test, the NPV increases and the PPV decreases as the prevalence of the variable being detected by the test decreases, whereas the NPV decreases and the PPV increases as the prevalence of the variable being detected by the test increases. The number of calves evaluated in the present study was fairly small (n = 28), and as expected, the apparent prevalence was low for all 3 bacterial organisms evaluated in both NPS and BALF samples collected 5 days after gamithromycin administration. Therefore, it was not surprising that, in general, the NPVs were higher and the PPVs were lower in this study.
In the present study, culture results for NPS samples were simply described because comparison of the results over time was not straightforward owing to the unknown effect of gamithromycin administration and small study population. However, it was interesting that within both the NPS and BALF samples, several calves with clinical signs of BRD were culture negative for a specific organism immediately prior to gamithromycin administration but subsequently became culture positive for that organism. This might have been the result of bacterial transmission among calves commingled in the same pen. It is also possible that a specific organism was present in the respiratory tract before gamithromycin administration but was not successfully cultured for various reasons such as competitive inhibition.10 Multiple calves had NPS samples that were either culture positive or negative for a specific organism on all sampling days. Consistent culture-positive samples might have been the result of organism transmission among calves. It is also possible that some calves require > 5 days after treatment to clear specific organisms owing to variable immune responses among individuals or resistance of the infecting bacterial strain to the antimicrobial used for treatment, which might preclude clearance of the organism. Consistent culture-negative samples might result from an absence of specific organisms or be reflective of the fairly low sensitivity of bacterial culture of NPS and BALF samples. Although interpretation of the bacterial culture results in this study was complicated by the timing of sample collection relative to gamithromycin administration, variability in culture results for serially collected samples from an individual animal is not a new finding. For example, in 1 study,9 isolation of M haemolytica and P multocida was quite variable between the left and right nasal cavities of calves when they were serially sampled multiple times per day for 5 days. In another study,10 the apparent prevalence of M haemolytica was inconsistent when determined from culture results for unilateral NPS samples that were collected from a small group of healthy calves once daily for 3 days. Collectively, these results indicate that there is a dynamic relationship between bacterial organisms and individual calves and suggest that a negative or positive culture result for a specific organism in a sample collected at a given time does not mean that the same culture result will be yielded by a similar sample collected at a different time.
Evaluation of the gamithromyin susceptibility profiles was limited to descriptive analysis because of the small study population and even smaller number of paired NPS and BALF samples in which both samples cultured positive for a specific organism. The gamithromycin susceptibility phenotype (susceptible, intermediate, or resistant) was generally in agreement for paired NPS and BALF samples in which both samples cultured positive for a specific organism. In fact, of the 56 paired NPS and BALF samples evaluated, the gamithromycin susceptibility phenotype was in agreement for all 4 pairs in which both samples cultured positive for H somni; however, that apparent perfect agreement may simply be an artifact of the low number of H somni culture-positive samples. For M haemolytica and P multocida, there were several instances in which either an NPS or BALF sample contained isolates with different gamithromycin susceptibility profiles, but only 1 susceptibility phenotype was obtained from isolates cultured from the other sample type that was collected at the same time. On the basis of the results of this study, it would seem that the gamithromycin susceptibility phenotype for P multocida isolates from paired NPS and BALF samples agreed more frequently than that for M haemolytica isolates. However, the gamithromycin susceptibility was determined for up to 12 M haemolytica isolates but only ≤ 6 P multocida isolates from each culture plate. Thus, for each sample, there were fewer opportunities to identify P multocida isolates with varying gamithromycin susceptibility profiles than to identify M haemolytica isolates with varying gamithromycin susceptibility profiles. Nevertheless, the results of the present study indicated that, although rare, it is possible for > 1 antimicrobial susceptibility phenotype to exist within genera and species of bacteria isolated from 1 sample, and the antimicrobial susceptibility phenotype for a specific bacterial organism cultured from the nasopharynx may differ from that for the same bacterial organism cultured from the lung. It is important to note that, in this study, the gamithromycin susceptibility was determined for multiple isolates of each bacterial species cultured on the same plate, a practice that is not commonly performed in routine antimicrobial susceptibility testing. Therefore, in routine diagnostic settings, it is unlikely to get mixed antimicrobial susceptibility results for a bacterial organism cultured from 1 sample.
Although identifying > 1 genetic subtype of M haemolytica in 1 URT or LRT sample was rare in the present study, the overall number of M haemolytica culture-positive samples was fairly low (32/131 [24%]); therefore, the frequency of this phenomenon in another population with a higher prevalence of M haemolytica may differ. Additionally, on a given plate, no more than 12 isolates underwent genomic sequencing regardless of the number of suspect M haemolytica colonies present, so it is possible that some genetic subtypes within a sample were not identified. Given the limitations of current bacterial culture methods and the dynamic microbial ecology within the respiratory tract of individual calves, it is also possible that additional M haemolytica subtypes were present but not cultured. Other possible explanations for failure to identify > 1 genetic subtype within a sample include competitive inhibition between certain M haemolytica subtypes and other microbes, selective inhibition of certain M haemolytica subtypes by the immune system or gamithromycin, selection bias of isolates in the laboratory, and the presence of a true genetic subtype monoculture. Similar to evaluation of the gamithromycin susceptibility profiles, assessment of whole-genome sequencing results for M haemolytica was limited to descriptive analysis because of the small study population and low number of M haemolytica culture-positive samples. Therefore, the only conclusion that can be drawn from these results is that different genetic subtypes of M haemolytica can be identified from 1 NPS sample when up to 12 isolates are examined. Additionally, when the gamithromycin susceptibility results were compared with the genomic sequencing results for M haemolytica, it became apparent that gamithromycin susceptibility varied within the same subtype. This suggested that, although isolates of the same subtype are genetically similar, they are not genetically identical.
Interpretation of diagnostic test results for calves with BRD is challenging, even with currently available technology and knowledge. Regardless of the method used to obtain respiratory tract samples from live calves, each technique is limited by the fairly small area that is sampled,10 potentially complex interactions among microbial flora that may be present in some calves but not others, and the inherent limitations of current bacterial culture methods.19 Additionally, the ability to discriminate between calves with subclinical and clinical disease is often suboptimal,20 and the likelihood of isolating target pathogens such as M haemolytica, P multocida, and H somni may differ as the stage of BRD varies. Although bacterial organisms cultured from samples obtained by NPSs and BAL may not be representative of the microbial flora of the entire URT or LRT5 and discrepancies in the culture results and antimicrobial susceptibility profiles can occur, there are few other alternatives for the collection of respiratory tract samples from live cattle in a field setting. Nevertheless, it is critical that the limitations of each diagnostic test are carefully considered during interpretation of test results, especially when antimicrobial susceptibility testing is performed and the results are used to guide treatment and management decisions.
In the present study, predictive values and κ values for the comparison of culture results between paired NPS and BALF samples were calculated only for samples collected on day 5; however, this information regarding the comparison of bacterial culture results between samples obtained from the URT and LRT of calves following antimicrobial treatment is valuable to the industry. In both clinical and field settings, samples are frequently not obtained for bacterial culture and antimicrobial susceptibility testing prior to antimicrobial administration and instead are collected only from calves that fail to respond to treatment or die and are necropsied. Additionally, mass medication of calves with an antimicrobial at feedlot arrival is a common practice; therefore, many calves that subsequently develop clinical BRD will have had some exposure to an antimicrobial prior to treatment for the disease. Because of the aforementioned knowledge gaps regarding bacteria involved in the pathogenesis of BRD and the fact that the PPVs and NPVs of diagnostic tests vary with disease prevalence, information regarding diagnostic test performance (ie, sensitivity, specificity, PPV, and NPV) and the bacterial organisms isolated from the respiratory tract of calves with BRD at various times before, during, and after antimicrobial treatment will be beneficial for the interpretation of test results, and further research in this area is necessary.
Results of the present study indicated that there was a high level of agreement between the M haemolytica, P multocida, and H somni culture results obtained for concurrently collected NPS and BALF samples from calves with BRD 5 days after treatment with gamithromycin. However, that high level of agreement and the generally high NPV of bacterial culture of NPS specimens relative to BALF samples were likely influenced by the low number of culture-positive samples and may differ in other populations. The culture results for both NPS and BALF samples indicated that bacterial organisms isolated within an individual calf can vary over time. The results also indicated that a calf with clinical BRD can have different genetic subtypes of M haemolytica present in the nasopharynx at the same time, a mixture of gamithromycin susceptibility phenotypes for a specific organism can be present within 1 NPS or BALF sample, and paired NPS and BALF samples may have different gamithromycin susceptibility phenotypes for a specific organism when multiple isolates are tested from each sample. Therefore, bacterial culture results for NPS and BALF samples obtained from beef feedlot calves with clinical BRD should be interpreted with caution following consideration of the expected prevalence of the disease within the population, timing of the sample collection relative to antimicrobial administration, and inherent limitations of currently available bacterial culture and antimicrobial susceptibility testing methods.
Acknowledgments
Supported by Merial, the USDA Agricultural Research Service, and in part by the Beef Checkoff Program. Merial was not involved in any aspect of designing the study, conducting the microbiological analyses, interpreting the data, or preparing the manuscript.
The authors declare that there were no conflicts of interest.
The authors thank Max Andersen, Kelly Lechtenberg, Sara McReynolds, Bradley Robért, Kelsey McClure, and Mal Hoover for technical assistance.
ABBREVIATIONS
| BAL | Bronchoalveolar lavage |
| BALF | Bronchoalveolar lavage fluid |
| BRD | Bovine respiratory disease |
| LRT | Lower respiratory tract |
| NPS | Nasopharyngeal swab |
| NPV | Negative predictive value |
| PPV | Positive predictive value |
| SNP | Single-nucleotide polymorphism |
| URT | Upper respiratory tract |
Footnotes
VetOne, Boise, Idaho.
SurgiVet, Smiths Medical, Dublin, Ohio.
Sensititre, TREK Diagnostic Systems, Thermo Fisher Scientific Inc, Waltham, Mass.
Ultra Clean-htp 96-Well Microbial DNA Isolation Kit, Mo Bio Laboratories Inc, Carlsbad, Calif.
Quantus fluorometer, Promega Corp, Madison, Wis.
Nextera XT DNA libraries and original A Indices kits, Illumina Inc, San Diego, Calif.
MiSeq Sequencer, Illumina Inc, San Diego, Calif.
Langmead B, Trapnell C, Pop M, et al. Ultrafast and memory efficient alignment of short DNA sequences to the human genome (abstr). Genome Biol 2009;10:R25.
PHYLIP, version 3.69, Felsenstein J, Department of Genetics, University of Washington, Seattle, Wash. (Distributed by the author, 1993).
Dendroscope, version 3.2.10, Huson DH, MathematischNaturwissenschaftliche Fakultät, Universität Tübingen, Tübingen, Germany.
STATA, version 12.1, StataCorp LP, College Station, Tex.
References
1. Centers for Epidemiology and Animal Health. Types and costs of respiratory disease treatments in US feedlots. Fort Collins, Colo: USDA APHIS Veterinary Services Centers for Epidemiology and Animal Health, 2013. Available at: www.aphis.usda.gov/animal_health/nahms/feedlot/downloads/feedlot2011/Feed11_is_RespDis.pdf. Accessed Sep 29, 2016.
2. Griffin D, Chengappa MM, Kuszak J, et al. Bacterial pathogens of the bovine respiratory disease complex. Vet Clin North Am Food Anim Pract 2010; 26:381–394.
3. DeDonder KD, Apley MD, Li M, et al. Pharmacokinetics and pharmacodynamics of gamithromycin in pulmonary epithelial lining fluid in naturally occurring bovine respiratory disease in multisource commingled feedlot cattle. J Vet Pharmacol Ther 2016; 39:157–166.
4. DeRosa DC, Mechor GD, Staats JJ, et al. Comparison of Pasteurella spp simultaneously isolated from nasal and transtracheal swabs from cattle with clinical signs of bovine respiratory disease. J Clin Microbiol 2000; 38:327–332.
5. Allen JW, Viel L, Bateman KG, et al. The microbial flora of the respiratory tract in feedlot calves: associations between nasopharyngeal and bronchoalveolar lavage cultures. Can J Vet Res 1991; 55:341–346.
6. Godinho KS, Sarasola P, Renoult E, et al. Use of deep nasopharyngeal swabs as a predictive diagnostic method for natural respiratory infections in calves. Vet Rec 2007; 160:22–25.
7. Bendall D. Clinical case study: use of bronchial alveolar lavage (BAL) to investigate a pneumonia outbreak. UK Vet Livestock 2007; 12:51–52.
8. Caldow G. Bronchoalveolar lavage in the investigation of bovine respiratory disease. In Pract 2001; 23:41–43.
9. Magwood SE, Barnum DA, Thomson RG. Nasal bacterial flora of calves in healthy and in pneumonia-prone herds. Can J Comp Med 1969; 33:237–243.
10. Capik SF, White BJ, Lubbers BV, et al. Characterization of Mannheimia haemolytica in beef calves via nasopharyngeal culture and pulsed-field gel electrophoresis. J Vet Diagn Invest 2015; 27:568–575.
11. Clinical and Laboratory Standards Institute. VET01-A4: Performance standards for antimicrobial disk and dilution susceptibility tests for bacteria isolated from animals; approved standard, fourth edition. Wayne, Pa: Clinical and Laboratory Standards Institute, 2013;38–39.
12. Clawson ML, Murray RW, Sweeney RW, et al. Genomic signatures of Mannheimia haemolytica that associate with the lungs of cattle with respiratory disease, an integrative conjugative element, and antibiotic resistance genes. BMC Genomics 2016; 17:982.
13. Harhay GP, Koren S, Phillippy AM, et al. Complete closed genome sequences of Mannheimia haemolytica serotypes A1 and A6, isolated from cattle. Genome Announc 2013; 1:e00188–e13.
14. Huson DH, Richter DC, Rausch C, et al. Dendroscope: an interactive viewer for large phylogenetic trees. BMC Bioinformatics 2007; 8:460.
15. Dohoo I, Martin W, Stryhn H. Screening and diagnostic tests. In: Veterinary epidemiologic research. 2nd ed. Charlottetown, PE, Canada: VER Inc, 2009;97–99.
16. Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics 1977; 33:159–174.
17. Vach W. The dependence of Cohen's kappa on the prevalence does not matter. J Clin Epidemiol 2005; 58:655–661.
18. Panciera RJ, Confer AW. Pathogenesis and pathology of bovine pneumonia. Vet Clin North Am Food Anim Pract 2010; 26:191–214.
19. Singer RS, Johnson WO, Jeffrey JS, et al. A statistical model for assessing sample size for bacterial colony selection: a case study of Escherichia coli and avian cellulitis. J Vet Diagn Invest 2000; 12:118–125.
20. Wittum TE, Woollen NE, Perino LJ, et al. Relationships among treatment for respiratory tract disease, pulmonary lesions evident at slaughter, and rate of weight gain in feedlot cattle. J Am Vet Med Assoc 1996; 209:814–818.
