|Click here for information on how to order reprints of this article.|
A Novel Cytologic Evaluation Technique for the Detection of Mycobacteria in Veterinary Samples
Kelly L. Diegel, DVM, PhD*
Scott D. Fitzgerald, DVM, PhD*
Mitchell V. Palmer, DVM, PhD†
Diana L. Whipple, MS†
*Diagnostic Center for Population and Animal Health, Michigan State University College of Veterinary Medicine, East Lansing, Michigan
†Bacterial Diseases of Livestock Research Unit, National Animal Disease Center, Agricultural Research Service, USDA, Ames, Iowa
KEY WORDS: Mycobacteria, cytology, acid-fast, diagnostic, veterinary, white-tailed deer, tuberculosis
Mycobacterial culture and identification is currently the “gold standard” technique for the diagnosis of mycobacteriosis in animals. Unfortunately, confirming infection using this technique can be time-consuming and cost-prohibitive in the veterinary setting, particularly when large numbers of samples require testing. The objective of this study was to compare results of a cytology-based procedure to culture results in experimentally infected deer. Cell-rich samples were collected from 19 white-tailed deer (Odocoileus virginianus) inoculated by intratonsilar instillation of 2¥108 colony forming units (CFU) of Mycobacterium bovis. These cell samples were processed using both a commercial, automated cytology device and by routine mycobacterial culture. Cytology was both less sensitive and less specific than culturing of samples for M. bovis tuberculosis, but it was also time- and cost-efficient. This cytologic method may have substantial value in species in which skin testing is unreliable for the antemortem diagnosis of tuberculosis.
Mycobacterium bovis tuberculosis recently regained attention in the state of Michigan, when an endemic strain of M. bovis tuberculosis established in free-ranging white-tailed deer became a major threat to livestock producers. This threat resulted in the loss of the tuberculosis-free status of the state. The need for efficient and reliable diagnostic testing for tuberculosis in deer has been heightened in the face of this endemic.
Currently, the gold standard diagnostic test for tuberculosis remains mycobacterial culture (isolation and identification), which, although sensitive and specific, often takes weeks to complete. Histopathologic examination of acid-fast stained tissue sections can be a very specific diagnostic tool, but it depends on gross suspicion or actual detection of lesions. Culture results are still typically necessary to confirm infection. Additionally problematic is that, in veterinary medicine, culture and histology are primarily postmortem analyses.
The widely accepted tuberculin (purified protein derivative, PPD) skin test varies in reported sensitivities and specificities but remains one of the best antemortem detection methods for the disease. Reported specificities and sensitivities for tuberculin testing, according to a review of the literature, vary from approximately 72% to over 99% in cattle, with an even wider range reported for deer.2,3 Results are often based on the type of skin test applied (single cervical, comparative cervical, or caudal fold), whether the result is used in series with other diagnostic techniques, the species on which the test is applied, and the geographic location of the cattle or deer herd.2,4
One simple and relatively inexpensive technique that is rarely used in diagnosing tuberculosis in animals is cytology. Cytologic evaluations of proposed shedding sites have not been investigated in animals. Because mycobacteria are intracellular organisms, if present in adequate numbers in cells near shedding sites, they should be detectable within cells harvested from these sites. The goal of this study was to investigate whether such detection was possible. Moreover, because knowledge of transmission routes of tuberculosis in white-tailed deer is limited, we also sought to better define potential shedding routes. We investigated the potential applications of a cytology preparation method previously used only in human medicine: a commercially-available automated cytology device (ThinPrep 2000; Cytyc Corp., Boxborough, MA). In human clinical diagnosis, the processing technique used in this study has reportedly resulted in more efficient diagnosis with fewer false negative results and a reduction in the preparation error present in conventional, direct-smear techniques.5–8 Cell quality and quantity available for diagnosis is reportedly greatly improved compared with traditional smear preparation methods.6,8 However, its application in the detection of mycobacteria in cell-rich samples has not been investigated.
Materials and Methods
Nineteen 6-month-old white-tailed deer (Odocoileus virginianus) were experimentally inoculated with M. bovis as part of a high-dose, short term inoculation study at the National Animal Disease Center (NADC) in Ames, Iowa. These 8 castrated males and 11 females underwent intratonsilar instillation of 2 ¥ 108 colony forming units (CFU).9 Deer were housed inside a biosecurity level 3 building with directional airflow such that air from the animal pens was pulled towards a central corridor and passed through high-efficiency particulate air (HEPA) filters before exiting the building. Airflow velocity was controlled to provide 10.4 air changes/min in the animal pens. Deer in each pen had access to a circulating watering device and were fed a complete pelleted feed for deer and elk (Complete Feed 55P3; Purina Mills, St. Louis, MO) and alfalfa hay. Pens were cleaned once daily, one at a time, by transferring deer to a holding pen and thoroughly washing the floor and lower walls of the empty pen with a high-pressure hose. During cleaning, deer had contact with pen-mates only and not with deer from other pens.
All deer in this study were tested by comparative cervical test (CCT) the month before inoculation and then again near the 3-month cytology/culture sample collection date as part of a larger study intended to assess usefulness of the CCT in white-tailed deer. Detailed methods and results of this separate study were previously published.4 Deer were injected in the midcervical area skin with 0.1 mL each avian and bovine PPD at each of the 2 testing dates. Injection sites were observed, palpated, and measured for thickness 72 hours after PPD injection.
Between days 21 and 63 after inoculation, 3 deer were euthanized due to injuries acquired during handling. Four deer were euthanized between 63 and 90 days after inoculation due to poor condition from advanced tuberculosis, and another 3 deer were euthanized due to injuries from handling. Between 90 and 113 days after inoculation, 1 deer was euthanized due to advanced tuberculosis and another due to injuries. One hundred twenty days after inoculation, the 7 remaining experimentally inoculated animals were euthanized. Consequently, as the study progressed, the number of animals available for sampling decreased (Table 1).
For inoculation and sampling, deer were anesthetized with a combination of xylazine (Mobay Corporation, Shawnee, KS) (2 mg/kg body weight) and ketamine (Fort Dodge Laboratories, Fort Dodge, IA) (6 mg/kg) injected intramuscularly. The effects of xylazine were reversed with tolazoline (Lloyd Laboratories, Shanandoah, IA) (4 mg/kg) injected intravenously.
The strain of M. bovis used was strain 1315, originally isolated from a free-ranging tuberculous white-tailed deer killed by a hunter in Alpena county, Michigan, in 1994. The isolate was incubated at 37˚C for 6 weeks on Middlebrook 7H9 liquid media with 10% oleic albumin dextrose citrate enrichment (Bacto Middlebrook OADC Enrichment; DIFCO Laboratories, Detroit MI). After incubation, the bacteria were harvested by centrifugation and washed twice with 0.01 mol phosphate buffered saline (PBS) solution, pH 7.4. After resuspension in PBS solution, serial 10-fold dilutions were inoculated on Middlebrook 7H10 agar slants supplemented with OADC to determine the number of CFU. Inoculum was then frozen at -80˚C for future use.
Nasal and oral swabs were collected for cytologic evaluation and bacteriologic culture on days 21, 63, 90, and 113 after inoculation. Swabs of the tonsilar crypts were collected 21 days after inoculation and at the time of necropsy. Swabs for cytologic evaluation and bacteriologic culture were collected using a sterile 18-cm cytology brush (Puritan Medical Products, Guilford, ME). For bacteriologic culture, swabs were rinsed thoroughly in 1.0 mL PBS. One half of the sample (0.5 mL) was added to 0.2% benzalkonium chloride solution (Zephiran chloride, concentrate 17%, Sterling Drug, New York, NY) and left at room temperature for 15 minutes. After decontamination with benzalkonium chloride, samples were centrifuged for 20 minutes at 2,000 rpm (750 ¥ g) and the supernatant was decanted. Then, 0.5 mL of Bacto egg yolk enrichment 50% (DICFO Laboratories) was added to the sediment. Samples of the sediment-egg yolk combination were inoculated onto separate agar slants containing Stonebrink’s, Harrold’s egg yolk, Middlebrook 7H10, or Middlebrook 7H11 media. Inoculated agar slants were incubated at 37˚C for 8 weeks.
Cytology Preparation and Scoring
Samples were obtained from the NADC research deer as described previously. Cytology brushes were placed in 20 mL liquid preservative (PreservCyt solution; Cytyc Corp.) contained in a 2 ounce plastic vial and were shipped to the Animal Health Diagnostic Laboratory (AHDL), Michigan State University (MSU), for preparation. Specimens were processed as recommended for mucoid samples according to the equipment operator’s manual (Cytyc Corp.) in a method previously described.10 Briefly, this procedure involves vortexing the collected brush sample, running the sample on the automated cytology machine, and staining the resulting cell preparation. While in the automated cytology device, sample fluid in the plastic vial is first rotated to disperse cell materials from debris. Cells are then collected from the fluid across a filtration membrane using vacuum pressure. Filtered cells are then evenly pressed to a glass slide by the machine, in a circular area of approximately 20 mm in diameter. In this study, prepared slides with adherent cells were then fixed in 95% ethanol for 10 minutes or longer and stained using a modification of the Ziehl-Neelsen technique for the identification of acid-fast bacteria in cytology preparations.11
Slides were prepared and evaluated before any culture result was reported and were read blindly using light microscopy. Subjective cellularity scores were assigned to each slide (poor, moderate, or good cellularity). Slides were graded for the presence of mycobacteria at ¥ 40 magnification. Grades were assigned as follows: less than 1 acid-fast organism noted per high power field (HPF; ¥ 40) = 1+; 2 to 10 organisms noted per HPF = 2+; more than 10 organisms per HPF = 3+.
Descriptive statistics (sensitivities, specificities, positive and negative predictive values) were calculated using a standard software package (Microsoft Excel software; Microsoft, Redmond, WA). These values were generated by comparing the results of cytology to those of culture swabs obtained from the same anatomic sites. A swab from which M. bovis was cultured was considered a true positive result. Values were calculated separately for each anatomic site and for the summed total of all results. Descriptive statistics comparing the comparative cervical test (CCT) results to the gold standard of culture were calculated using the same formulae. To avoid incalculable results for this comparison, a correction factor of 0.5 was added to all values.
Tonsil swabs could not be obtained for either culture or automated cytology sampling from any animals at 63 days after inoculation (Tables 2 and 3). Only 2 of 9 available animals could be sampled for cytologic evaluation at 90 days, due to anesthetic risks involved in their cases.
Skin test (CCT) results for these deer have been published elsewhere, as part of a larger study.4 No deer in this study were reactors before inoculation. At 3 months after inoculation, all deer but one were classified as reactors by CCT testing, even though many of these animals did not ultimately culture positive. The animal that was not a reactor at this point was considered suspect by CCT. This same animal had positive culture and cytology samples at 1 and 2 months after inoculation and positive cytology readings at the time the second skin test was performed (3 months after inoculation). When compared with culture at 90 days, CCT results had good sensitivity, but false positives were encountered (when culture is used as the gold standard comparison), resulting in a much lower specificity. Positive predictive value was poor at 35%, and negative predictive value was similarly poor at an only slightly higher 50% (Table 4).
The number of deer determined to be positive for M. bovis by both mycobacterial culture (Table 2) and by cytology (Table 3) are shown. Most samples that were positive had one or fewer organisms noted per HPF, a score of 1+ (Table 5). Eight of the 21 (38.1%) positive samples examined from the tonsil had a score of 2+, and 1 of 7 positive samples from the nasal cavity had a score of 2+ (14.3%). No samples received a 3+ grade for positivity. The new technique had an overall sensitivity of 73.17% and a specificity of 16.85 % when compared with the gold standard of mycobacterial culture (Table 6). Cytology preparations obtained from the tonsil had the highest sensitivity and specificity of any of the 3 sites calculated individually (84.21% and 55.56%, respectively). The overall positive predictive value for the test was 66.67%, and the overall negative predictive value was 87.06%. The site generating the highest individual positive predictive value was the nasal cavity (85.71%), whereas the highest negative predictive value resulted from samples obtained from oral cavity swabs (91.18%).
Subjective cellularity grades for the oral cavity were poorest. All samples from other sites had moderate to good cellularity reported for all slides, but samples from the oral cavity obtained at all dates after inoculation had poor scores. In this study, cellularity was quite good in all samples, except for some of the oral cavity brush swabs. Subjectively, all slide preparations were easily read and provided an adequate number of cells to base a positive or negative result. The stain procedure worked very well in highlighting organisms where present, and did not stain nonmycobacterial organisms (Figures 1–3).
The automated cytology device used in this study is currently Food and Drug Administration approved for use in diagnosis of multiple human samples, including but not limited to superficial scrapings, fluids, needle aspirates, mucoid samples (sputum, gastrointestinal), and gynecologic samples. To our knowledge, however, despite its wide range of potential clinical applications, its use for the diagnosis of mycobacterial disease has never been evaluated. Use of this technology for processing veterinary clinical or diagnostic samples has not been reported to date either, except for one reference mentioning its possible application in needle aspirate evaluation in research mice.12 In human literature, the technique has been evaluated to be as good as or better than cytocentrifuge or direct-smear preparations.10,13
The more important shedding routes, both by culture and cytology, seem to be the nasal and tonsilar routes, because oral samples were rarely positive by either technique. Pathophysiologically, these results seem reasonable, because tuberculosis is primarily a respiratory or lymphoid system-based disease in many large animal species. Because this is the first such study in deer, more data are necessary to confirm these findings, particularly because initial inoculation was intratonsilar in the study. Overall, of those sites assessed in this study, the oral route appears to be the least diagnostically rewarding in cases of tuberculosis. This is based on the low frequency of shedding and the relatively poor cellularity of oral cavity samples compared with tonsilar and nasal samples. It is possible that the sampling technique was responsible for much of the noted decreased cellularity in the oral samples.
The evaluation of cytology for diagnosing mycobacterial infection resulted in an overall sensitivity comparable to that reported for detecting malignancy in human specimens.10 Interestingly, the lack of specificity of the technique was the primary concern in that study as it appears to be in ours. The fact that the tonsil was the tissue with the highest sensitivity and specificity throughout the study was most reasonably due to the fact that inoculation of deer was intratonsilar and shedding was most apparent in this tissue whenever samples were taken. The high negative predictive value suggests that shedding not detected by our test will not be detected by the gold standard technique, mycobacterial isolation and identification, in most cases.
The number of animals used in this study was small, but the resultant values show that our technique may have some valid diagnostic application in the antemortem diagnosis of tuberculosis in deer. The difficulty in restraining captive cervidae for testing became apparent both in the repeated collection of samples necessary for this study, and in the adjunct CCT study described. Limiting the number of times animals require restraint to obtain an accurate diagnosis is, of course, warranted. This is quite possible when samples or readings for multiple tests can be taken at one time.
Results suggest that cytology is at least comparable to CCT statistically, if not somewhat more reliable and accurate, when culture (mycobacterial isolation and identification) is used as the gold standard. Although cytology appeared to “miss” positive cases in some instances, false positives were not nearly as common as they were using CCT in these animals. Overall, cytology was comparable to culture in detecting actual infected animals. It would therefore follow that this test could be beneficial if used in conjunction with the CCT in deer. Results from this technique, when used in conjunction with skin test results, would validate cases in which active shedding may be occurring, regardless of the certainty of a CCT result. The test could help distinguish active shedders from animals that may have established but latent infections. Furthermore, cytology may be able to help diagnose active disease in those species in which no reliable skin testing protocol exists.
The described procedure certainly surpasses culture for mycobacteria in cost-efficiency and timeliness of diagnosis. The estimated turn-around time for this test approximates 48 hours, from sample collection to reporting. The cost in our laboratory was $18 per sample, including technician labor costs. When compared with the expected 6 to 12 weeks necessary for mycobacterial culture and the cost of culturing a clinical sample, approximately $110 based on the charges we incurred through contractual laboratory work, clearly the cytology technique is more rapid and cost-efficient. There appear to be sacrifices in both specificity and sensitivity of results to some degree with the technique, and no mycobacterial speciation can be performed after cytology examination alone. The low specificity encountered using cytology may be a direct result of this problem, with atypical mycobacteria in samples resulting in false-positive readings. Nevertheless, when used as a screening process in cases with suspected active infections, particularly when large numbers of samples may need to be processed (eg, in a herd situation), the benefits of the test are readily apparent.
No test will match the sensitivity and specificity of the available postmortem methods of confirmation of mycobacterial infection. However, when used in conjunction with other available antemortem tests, the presented cytology-based analysis of veterinary samples may prove to be a quick and inexpensive tool for the evaluation of mycobacterial infection status in many instances.
1. Schmitt SM, Fitzgerald SD, Cooley TM, et al: Bovine tuberculosis in free-ranging white-tailed deer from Michigan. J Wildlife Dis 33:749–758, 1997.
2. O’Reilly LM: Tuberculin skin tests: sensitivity and specificity, in Thoen CO and Steele JH (eds): Mycobacterium Bovis Infection in Animals and Humans. Ames, IA, Iowa State University Press, 1995:85–92.
3. Palmer MV, Whipple DL, Olsen SC, Jacobson RH: Cell mediated and humoral immune responses of white-tailed deer experimentally infected with Mycobacterium bovis. Res Vet Sci 68:95–98, 2000.
4. Palmer MV, Whipple DL, Waters WR: Tuberculin skin testing in white-tailed deer (Odocoileus virginianus). J Vet Diagn Invest 13:530–533, 2001.
5. Austin RM, Ramzy I: Increased detection of epithelial cell abnormalities by liquid-based gynecologic cytology preparations: A review of accumulated data. Acta Cytol 42:178–184, 1998.
6. Hutchinson ML, Isenstein LM, Goodman A, et al: Homogenous sampling accounts for the increased diagnostic accuracy using the ThinPrep processor. Am J Clin Pathol 101:215–219, 1994.
7. Linder J, Zahniser D: ThinPrep Papanicolaou testing to reduce false-negative cervical cytology. Arch Pathol Lab Med 122:139–144, 1998.
8. Tezuka F, Oikawa H, Shuki H, Higashiiwai H: Diagnostic efficacy and validity of the ThinPrep method in cervical cytology. Acta Cytol 40: 513–518, 1996.
9. Palmer MV, Whipple DL, Olsen SC: Development of a model of natural infection with Mycobacterium bovis in white-tailed deer. J Wildlife Dis 35:450–457, 1999.
10. Wang HH, Sovie S, Trawinski G, et al: Automated cytology device processing of endoscopic brushing specimens. Am J Clin Pathol 105:163–167, 1996.
11. Prophet EB: Laboratory methods in histotechnology, in Mills B, Arrington JB, Sobin LH (eds): American Registry of Pathology. Washington, DC, Armed Forces Institute of Pathology, 1992.
12. Kobayashi TK, Nakano Y, Sugimoto H, et al: Cellular preparations of fine needle aspirates from transplanted tumour in nude mice using ThinPrep technique. Cytopath 8: 354–357, 1997.
13. Papillo JL, Lapen D: Cell yield: ThinPrep vs cytocentrifuge. Acta Cytol 38:33–36, 1994.
Table 1. Deer Numbers at Postinoculation Sampling Dates
Table 2. Deer That Cultured Positive for M. Bovis
Table 3. Deer With Positive Cytology Readings*
21 19/19 0/19 2/19
63 0/0 6/16 6/16
90 0/9 3/9 2/9
113 0/7 2/7 1/7
*Results are given as number of samples with listed score over total number of positive samples for listed tissue.
assigned to average reading for entire slide:
21 16/19 7/19 0/19
63 0/0 5/16 3/16
90 1/2 1/9 2/9
113 4/7 4/7 2/7
*Acid-fast bacilli noted with light microscopy.
Table 4. Comparison of CCT to Culture as Gold Standard 90 Days After Inoculation*
Table 5. Cytology scores for samples with positive readings*
Number of deer positive 9 3
Number of deer negative 0 6
Positive predictive value§** 35.00
Negative predictive value**¶ 50.00
*Deer considered suspect and reactors on CCT evaluation were considered positive. Deer with at least one swab site (oral, nasal, or tonsil) yielding M. bovis on culture/isolation were considered positive. Only 90 day after inoculaton, culture results are included in calculations.
†The number of true positive samples (percentage positive by cytology that were positive by culture).
‡The number of true negative samples (percent negative by cytology that were negative by culture).
§The probability that positive cytologies were truly positive when compared with the “gold standard.”
¶The probability that negative cytology samples were truly negative on culture.
**A value of 0.5 was added to each cell for descriptive statistical calculation, to prevent zero values within the numerator of equations used for these calculations that lead to incalculable results.
1+ 13/21 17/17 6/7
2+ 8/21 0/17 1/7
3+ 0/21 0/17 0/7
*Results are given as number of samples with listed score over total number of positive samples for listed tissue.
†Scores assigned to average reading for entire slide: 1+ = 1 or fewer organisms per high power field (HPF; 40X); 2+ = 2 to 10 bacilli per HPF; 3+ = more than 10 bacilli per HPF.
Table 6. Results of Descriptive Statistical Analysis of the Automated Cytology Device Technique When Compared With Culture as the Gold Standard
Sensitivity* 84.21 72.73 54.55 73.17
Specificity† 55.56 22.50 2.50 16.85
76.19 47.06 85.71 66.67
Negative 57.14 91.18 88.64 87.06
*The number of true positive samples (percentage positive by cytology that were positive by culture).
†The number of true negative samples (percent negative by cytology that were negative by culture).
‡The probability that positive cytologies were truly positive when compared with the “gold standard.”
§The probability that negative cytology samples were truly negative on culture.
Figure 1. Positive (2+) cytology obtained using the automated cytology device, tonsilar sample. Intracellular acid-fast bacilli (arrows). New Fuchsin stain; bar =10 µm.
Figure 2. Positive cytology (2+) obtained using the automated cytology device technique, nasal sample. Large group of intracellular acid-fast bacilli (arrow). New Fuchsin stain, bar = 10 µm.
Figure 3. Cytology example with large numbers of intracellular cocci and no acid-fast bacilli noted, automated cytology device technique. Photomicrograph displays clarity of cellular preparation for bacteria other than mycobacteria. New Fuchsin stain, bar = 10 µm.
|©2000-2010. All Rights Reserved. Veterinary Solutions LLC