Use of whole-genome sequencing and evaluation of the apparent sensitivity and specificity of antemortem tuberculosis tests in the investigation of an unusual outbreak of Mycobacterium bovis infection in a Michigan dairy herd

Colleen S. Bruning-Fann Veterinary Services, APHIS, USDA, 3001 Coolidge Rd #325, East Lansing, MI 48823.

Search for other papers by Colleen S. Bruning-Fann in
Current site
Google Scholar
PubMed
Close
 DVM, MS
,
Suelee Robbe-Austerman National Veterinary Services Laboratories, APHIS, USDA, 1920 Dayton Ave, Ames, IA, 50010.

Search for other papers by Suelee Robbe-Austerman in
Current site
Google Scholar
PubMed
Close
 DVM, PhD
,
John B. Kaneene Center for Comparative Epidemiology, College of Veterinary Medicine, Michigan State University, East Lansing, MI 48824.

Search for other papers by John B. Kaneene in
Current site
Google Scholar
PubMed
Close
 DVM, MPH, PhD
,
Bruce V. Thomsen National Veterinary Services Laboratories, APHIS, USDA, 1920 Dayton Ave, Ames, IA, 50010.

Search for other papers by Bruce V. Thomsen in
Current site
Google Scholar
PubMed
Close
 DVM, PhD
,
John D. Tilden Jr Food and Dairy Divisions, Michigan Department of Agriculture and Rural Development, 525 W Allegan St, Lansing, MI 48933.

Search for other papers by John D. Tilden Jr in
Current site
Google Scholar
PubMed
Close
 DVM, MPH
,
Jean S. Ray Veterinary Services, APHIS, USDA, 3001 Coolidge Rd #325, East Lansing, MI 48823.

Search for other papers by Jean S. Ray in
Current site
Google Scholar
PubMed
Close
 DVM, PhD
,
Richard W. Smith Animal Industry Divisions, Michigan Department of Agriculture and Rural Development, 525 W Allegan St, Lansing, MI 48933.

Search for other papers by Richard W. Smith in
Current site
Google Scholar
PubMed
Close
 DVM
,
Scott D. Fitzgerald Diagnostic Center for Population and Animal Health, College of Veterinary Medicine, Michigan State University, Lansing, MI 48910.

Search for other papers by Scott D. Fitzgerald in
Current site
Google Scholar
PubMed
Close
 DVM, PhD
,
Steven R. Bolin Diagnostic Center for Population and Animal Health, College of Veterinary Medicine, Michigan State University, Lansing, MI 48910.

Search for other papers by Steven R. Bolin in
Current site
Google Scholar
PubMed
Close
 DVM, PhD
,
Daniel J. O'Brien Wildlife Disease Laboratory, Michigan Department of Natural Resources, 4125 Beaumont Rd, Lansing, MI 48910.

Search for other papers by Daniel J. O'Brien in
Current site
Google Scholar
PubMed
Close
 DVM, PhD
,
Thomas P. Mullaney Diagnostic Center for Population and Animal Health, College of Veterinary Medicine, Michigan State University, Lansing, MI 48910.

Search for other papers by Thomas P. Mullaney in
Current site
Google Scholar
PubMed
Close
 MVB, PhD
,
Tod P. Stuber National Veterinary Services Laboratories, APHIS, USDA, 1920 Dayton Ave, Ames, IA, 50010.

Search for other papers by Tod P. Stuber in
Current site
Google Scholar
PubMed
Close
 BS
,
James J. Averill Animal Industry Divisions, Michigan Department of Agriculture and Rural Development, 525 W Allegan St, Lansing, MI 48933.

Search for other papers by James J. Averill in
Current site
Google Scholar
PubMed
Close
 DVM, PhD
, and
David Marks Wildlife Services, APHIS, USDA, 2651 Coolidge Rd #101, East Lansing, 48823.

Search for other papers by David Marks in
Current site
Google Scholar
PubMed
Close
 MS

Click on author name to view affiliation information

Abstract

OBJECTIVE To describe use of whole-genome sequencing (WGS) and evaluate the apparent sensitivity and specificity of antemortem tuberculosis tests during investigation of an unusual outbreak of Mycobacterium bovis infection in a Michigan dairy herd.

DESIGN Bovine tuberculosis (bTB) outbreak investigation.

ANIMALS Cattle, cats, dog, and wildlife.

PROCEDURES All cattle in the index dairy herd were screened for bTB with the caudal fold test (CFT), and cattle ≥ 6 months old were also screened with a γ-interferon (γIFN) assay. The index herd was depopulated along with all barn cats and a dog that were fed unpasteurized milk from the herd. Select isolates from M bovis–infected animals from the index herd and other bTB-affected herds underwent WGS. Wildlife around all affected premises was examined for bTB.

RESULTS No evidence of bTB was found in any wildlife examined. Within the index herd, 53 of 451 (11.8%) cattle and 12 of 21 (57%) cats were confirmed to be infected with M bovis. Prevalence of M bovis–infected cattle was greatest among 4- to 7-month-old calves (16/49 [33%]) followed by adult cows (36/203 [18%]). The apparent sensitivity and specificity were 86.8% and 92.7% for the CFT and 80.4% and 96.5% for the γIFN assay when results for those tests were interpreted separately and 96.1% and 91.7% when results were interpreted in parallel. Results of WGS revealed that M bovis–infected barn cats and cattle from the index herd and 6 beef operations were infected with the same strain of M bovis. Of the 6 bTB-affected beef operations identified during the investigation, 3 were linked to the index herd only by WGS results; there was no record of movement of livestock or waste milk from the index herd to those operations.

CONCLUSIONS AND CLINICAL RELEVANCE Whole-genome sequencing enhanced the epidemiological investigation and should be used in all disease investigations. Performing the CFT and γIFN assay in parallel improved the antemortem ability to detect M bovis–infected animals. Contact with M bovis–infected cattle and contaminated milk were major risk factors for transmission of bTB within and between herds of this outbreak.

Abstract

OBJECTIVE To describe use of whole-genome sequencing (WGS) and evaluate the apparent sensitivity and specificity of antemortem tuberculosis tests during investigation of an unusual outbreak of Mycobacterium bovis infection in a Michigan dairy herd.

DESIGN Bovine tuberculosis (bTB) outbreak investigation.

ANIMALS Cattle, cats, dog, and wildlife.

PROCEDURES All cattle in the index dairy herd were screened for bTB with the caudal fold test (CFT), and cattle ≥ 6 months old were also screened with a γ-interferon (γIFN) assay. The index herd was depopulated along with all barn cats and a dog that were fed unpasteurized milk from the herd. Select isolates from M bovis–infected animals from the index herd and other bTB-affected herds underwent WGS. Wildlife around all affected premises was examined for bTB.

RESULTS No evidence of bTB was found in any wildlife examined. Within the index herd, 53 of 451 (11.8%) cattle and 12 of 21 (57%) cats were confirmed to be infected with M bovis. Prevalence of M bovis–infected cattle was greatest among 4- to 7-month-old calves (16/49 [33%]) followed by adult cows (36/203 [18%]). The apparent sensitivity and specificity were 86.8% and 92.7% for the CFT and 80.4% and 96.5% for the γIFN assay when results for those tests were interpreted separately and 96.1% and 91.7% when results were interpreted in parallel. Results of WGS revealed that M bovis–infected barn cats and cattle from the index herd and 6 beef operations were infected with the same strain of M bovis. Of the 6 bTB-affected beef operations identified during the investigation, 3 were linked to the index herd only by WGS results; there was no record of movement of livestock or waste milk from the index herd to those operations.

CONCLUSIONS AND CLINICAL RELEVANCE Whole-genome sequencing enhanced the epidemiological investigation and should be used in all disease investigations. Performing the CFT and γIFN assay in parallel improved the antemortem ability to detect M bovis–infected animals. Contact with M bovis–infected cattle and contaminated milk were major risk factors for transmission of bTB within and between herds of this outbreak.

The resurgence of bTB in Michigan was discovered with the identification of a Mycobacterium bovis reservoir in wild WTD (Odocoileus virginianus) in the northeastern portion of the lower peninsula in 19951 and with the subsequent spillover of the disease into cattle2 and captive cervids.3 In response, livestock in that portion of the state became subject to movement restrictions, and extensive statewide surveillance testing of livestock for bTB began. Statewide testing of livestock for bTB was completed in 2003, with no bTB-affected herds identified outside the northern portion of the lower peninsula (bTB-endemic area). Since then, bTB surveillance in Michigan has been concentrated on livestock and wildlife within the bTB-endemic area. To date, 52 beef herds, 15 dairy herds, 4 beef feedlots, 4 captive cervid herds, and 1 bison herd have been identified as affected with bTB in Michigan, and the disease has cost millions of dollars in surveillance and the testing and removal of infected, exposed, and test-positive animals. Most M bovis–infected animals were in the early stages of the disease when they were identified, which limited transmission of the disease to other animals within and between populations.

The purpose of the report presented here was to describe the epidemiological investigation of a bTB outbreak that was traced back to a Michigan dairy farm located in Saginaw County (index herd) approximately 185 km (115 miles) south of the area where bTB is endemic in wildlife (Figure 1). This outbreak was unusual because, within the index herd, the prevalence of M bovis–infected animals varied markedly among age cohorts, and all herds involved in the outbreak were geographically located a considerable distance outside of the bTB-endemic area. Also, WGS was integral to the epidemiological investigation, and the testing protocol used allowed calculation of the apparent sensitivity and specificity of 2 antemortem bTB tests, the CFT and γIFN assay, when the tests were conducted individually and in parallel.

Figure 1—
Figure 1—

County map of the lower peninsula of Michigan that depicts the locations where bTB-infected deer (red dots; n = 794) and bTB-affected beef (blue triangles; 50) and dairy (yellow stars; 15) herds have been identified between 1975 and 2014. Beef and dairy herd symbols enclosed within red circles represent the 6 herds associated with the 2013 bTB outbreak investigation described in the present report that was traced back to a dairy farm located in Saginaw County (index herd) approximately 185 km (115 miles) south of the area where bTB is endemic in wildlife.

Citation: Journal of the American Veterinary Medical Association 251, 2; 10.2460/javma.251.2.206

Index Herd History

In February 2013, lesions consistent with bTB were detected in a cull dairy cow (index cow) during routine slaughter surveillance. That cow was traced back to a dairy farm (index herd) located in Saginaw County, Michigan, and an epidemiological investigation was initiated. The index herd was quarantined, tested, and confirmed to be infected with M bovis on March 25, 2013.

The index herd originated in 1985 with the purchase of 4 cows from a neighboring farm. During the 1990s, additional cows were purchased from farms located in the northeastern portion of the lower peninsula of Michigan before bTB was determined to be endemic in that area. By 2013, the herd had increased in size to 451 animals mainly through the retention of female cattle. A seasonal breeding program was used such that calves were born only during the months of March through December, with approximately half of the calves born from June through August. Cows were milked twice daily, and the herd had a rolling herd average milk production of 5,897 kg/cow/y (12,973 lb/cow/y). Prior to the detection of bTB in the index cow in 2013, the only bTB herd test was performed by a private practitioner in April 2001 as part of the statewide bTB surveillance program after identification of the M bovis reservoir in wildlife. During that herd test, all cattle > 18 months old were screened for bTB with the CFT. One hundred nine cattle were tested, with 4 classified as suspects; all 4 of those cattle subsequently tested negative for M bovis on the basis of results of a comparative cervical bTB test.

Epidemiological Investigation Protocol

Index herd testing procedures

At the time of the epidemiological investigation, the index herd consisted of 451 Holstein cattle with ages ranging from 4 to 78 months (mean, 29.6 months; median, 24 months), of which 375 were ≥ 12 months old. In accordance with the USDA bTB Eradication Uniform Methods and Rules,4 all cattle in the index herd were initially tested with the CFT. To increase the likelihood of identifying M bovis–infected cattle and to compare the efficacy of the CFT with the γIFN assay, all cattle ≥ 6 months old were tested with the CFT and γIFN assaya in parallel. The blood sample for the γIFN assay was collected on the day that the CFT was read (ie, 3 days after injection of the bovine purified protein derivative tuberculin). Cattle with a positive result on either the CFT or γIFN assay were classified as test-positive animals. All cattle in the index herd were euthanized or slaughtered regardless of test status, then necropsied or otherwise examined for evidence of M bovis infection. The owner received indemnification for all cattle that were necropsied.

Initially, the 9 cows with the greatest CFT responses and 9 young stock with the highest γIFN assay results were sent to the Michigan State University DCPAH for necropsy. From each animal, all suspicious lesions and specified lymph nodes (parotid; mandibular; medial retropharyngeal; cranial, middle, and caudal mediastinal; tracheobronchial; hepatic; cecal; mesenteric; and mammary lymph nodes)5 were collected, examined grossly, and sent for histologic examination and mycobacterial culture at the USDA NVSL.

To reduce costs, the remaining cattle in the herd were handled on the basis of their antemortem bTB test results and age. Test-positive calves (≤ 11 months old) were necropsied at the DCPAH with tissues collected and examined as previously described. Test-positive cattle > 12 months old were slaughtered at an abattoir where state or federal regulatory veterinarians collected lesions and specified lymph nodes, which were sent to the NVSL for histologic examination and mycobacterial culture. All test-negative cattle were slaughtered at the abattoir under the supervision of USDA Food and Safety Inspection Service veterinarians, who collected and submitted any granulomatous lesions characteristic of bTB to the NVSL for histologic examination and mycobacterial culture.

Tissue examination and mycobacterial culture

Formalin-fixed, paraffin-embedded tissues that contained both acid-fast bacilli and granulomatous lesions compatible with mycobacteriosis were evaluated by a PCR assay that used primers for IS 6110, which identify Mycobacterium tuberculosis–complex species. Samples were considered positive when appropriately sized amplicons were produced and observed after gel electrophoresis.6

All lesions and specified lymph nodes from test-positive animals, animals with gross bTB lesions, and milk samples collected from 19 test-positive cows were cultured for mycobacteria at the NVSL. For each test-positive animal, lymph nodes from each body region (head, thorax, and abdomen) were pooled together for culture. Tissue specimens were decontaminated by use of NaOH as described.7 Milk samples were decontaminated with an N-acetyl-l-cysteine procedure developed for culture of Mycobacterium avium subsp paratuberculosis as described8 and inoculated into an indicator mediumb and in-house Middlebrook 7H11 medium supplemented with pyruvate, hemolyzed blood, calf serum, and malachite green. Atypical mycobacteria were identified by use of 6S rRNA and rpoB partial sequencing as described.9 Cows were confirmed to be infected with M bovis on the basis of identification of the organism in tissue specimens or milk samples by PCR assay or mycobacterial culture.

WGS

The first 10 mycobacterial culture–positive cows and all animals with granulomatous lesions at multiple body regions or granulomatous lesions > 3 mm in size underwent WGS. For animals with mycobacterial culture–positive tissues at multiple body regions, tissues from at least 2 regions were sequenced to evaluate the extent of genetic diversity among mycobacterial isolates within the same animal. To avoid potential laboratory or selection bias, multiple primary colonies recovered from pooled tissue specimens from each of the 2 selected body regions were used to prepare the DNA.

Libraries were prepared by use of DNA library preparation and sequencing kits.c Files from the instrument were analyzed with the NVSL's in-house bioinformatics pipeline.d Briefly, reads were aligned to the M bovis reference AF2122/97 (NC_002945) by use of open-source softwaree as described.10,11 The files were processed with a best-practice workflow. Sequence files were deposited in the National Center for Biotechnology Information Sequence Read Archive under the bioproject PRJNA251692 (Supplementary Table S1, available at http://avmajournals.avma.org/doi/suppl/10.2460/javma.251.2.206).

Single nucleotide polymorphisms were called by use of open-source softwaree with ploidy set to 2 outputting SNPs to variant call files as described.12–14 Mycobacterial PPE protein–polymorphic GC-rich sequences and repeat regions were filtered, as were SNPs with a sequence quality score > 150 and the occasional homoplastic SNP (Supplementary Table S2, available at http://avmajournals.avma.org/doi/suppl/10.2460/javma.251.2.206). Single nucleotide polymorphisms were called as if processing a diploid organism (default ploidy set to 2). When ploidy was set to 2, the AC was reported as either 1 or 2 for each called SNP. If aligned reads had multiple numbers of reference and alternative calls at a position, the software reported the AC as 1, which represented only 1 allele different from the reference or a heterogeneous call for a haploid organism. Conversely, when the vast majority of calls were the alternative, the software reported the AC as 2, which represented 2 alleles different from the reference or, for a haploid organism, a consensus call in all reads suggestive of a single genotype.

To determine whether a mixed isolate was present, the total number of high-quality positions was determined by comparison of closely related isolates. If a sample had an AC of 1 at a high-quality position (as determined by evaluation of closely related isolates), that sample would be identified as a possible mixed infection in accordance with guidelines established for incompletely specified bases.15 To investigate the diversity of M bovis isolates within individual animals, isolates were sequenced from cattle in which the organism was cultured from multiple body regions or from tissue and milk. Results were reported in 3 formats: an aligned file, a tab-delimited file with the position location and SNPs grouped and sorted, and a phylogenetic tree created by use of maximum likelihood-based phylogenetic inference software.16 Finally parsimonious informative SNPs were validated visually by use of visualization software as described.17

Evaluation of cats and dog present on the index farm

Unpasteurized milk from cows in the index herd had been fed to 21 barn cats and available to 1 dog that resided on the farm. Those 22 animals were euthanized and necropsied at the DCPAH. All granulomatous lesions and specified lymph nodes were collected and underwent gross and histologic examination and mycobacterial culture as described.18

Other bTB surveillance

Cattle—Movement of all cattle and cattle products potentially contaminated with M bovis (eg, unpasteurized milk not saleable for human consumption [waste milk]) to and from the index herd was traced. All cattle at premises other than terminal feedlots (which send all cattle directly to slaughter) that received animals or milk from the index herd or had cattle that were exposed to cattle or milk from the index herd underwent a CFT. Additionally, all 37 cattle herds located within a 16-km (10-mile) radius of the index herd were screened for bTB as required by the memorandum of understanding19 between the state of Michigan and USDA for maintenance of its split-state (Modified-Accredited and Accredited-Free) status in regard to bTB.

Wildlife—The Michigan Department of Natural Resources and USDA Wildlife Services conducted surveillance of wild WTD in the townships (96.0 km2 [60.0 miles2]) around all bTB-affected premises associated with the index herd. Briefly, the formula used by the USDA Wildlife Services to select the geographic area for surveillance in wildlife is as follows: surveillance radius =

article image
, where A is the home range (in square miles) for the species in question.20 The estimated home range of wild WTD in Michigan is approximately 3.2 km2 (2.0 miles2).21 Deer were culled on privately owned land with landowner permission, and the resulting carcasses were examined for evidence of bTB as described.22,23 Additionally, deer harvested by hunters in Saginaw County and the other counties where bTB-affected premises associated with the index herd were located were likewise examined for evidence of bTB during the fall hunting seasons of 2013, 2014, and 2015.

The USDA Wildlife Services also set traps on the index herd property to capture common small wild mammals for bTB surveillance. Target species included raccoon (Procyon lotor), opossum (Didelphis virginiana), and coyote (Canis latrans), all of which can become infected with bTB.24–26

Results

Index herd testing

Of the 451 cattle in the index herd, 53 (11.8%) were confirmed to be infected with M bovis (Table 1) on the basis of identification of the organism in tissue specimens or milk by either a PCR assay or mycobacterial culture. All 451 cattle were screened for bTB with the CFT, and the 423 cattle ≥ 6 months old were also screened with the γIFN assay. Seventy-five of 451 (16.6%) cattle tested positive on the CFT, and 54 of 423 (12.8%) cattle tested positive on the γIFN assay. Of the 423 cattle that underwent both the CFT and γIFN assay, 49 (11.6%) had positive results for both tests, 26 (6.1%) had positive results for the CFT only, and 5 (1.2%) had positive results for the γIFN assay only. Overall, 80 (17.7%) cattle tested positive on the CFT or γIFN assay or both tests. Two of the 53 cattle that were confirmed to be infected with M bovis had negative results on both the CFT and γIFN assay.

Table 1—

Caudal fold test and γIFN assay results for the 451 Holsteins of a dairy herd (index herd) in Saginaw County, Michigan, that was depopulated and whose cattle were or were not subsequently confirmed to be infected with Mycobacterium bovis on the basis of identification of the organism in tissue specimens or milk by either a PCR assay or mycobacterial culture during a 2013 bTB outbreak investigation.

  M bovis infection status
CFT resultγIFN assay resultPositiveNegativeTotal
PositivePositive38 (8.4)11 (2.4)49 (10.9)
PositiveNegative8 (1.8)18 (4.0)26 (5.8)
NegativePositive3 (0.7)2 (0.4)5 (1.1)
NegativeNot performed2 (0.4)26 (5.8)28 (6.2)
NegativeNegative2 (0.4)341 (75.6)343 (76.0)
Total 53 (11.8)398 (88.2)451 (100)

Values represent the number (percentage) of animals. The γIFN assay was not performed on calves < 6 months old.

Mycobacterium bovis was cultured from the milk of 2 of the 19 index herd cows sampled. One of those cows had positive results for both the CFT and γ-IFN assay, whereas the other cow had positive results for the γIFN assay only. Interestingly, no gross or histologic lesions were observed in either cow, and unfortunately, tissue specimens from those 2 cows were not available for mycobacterial culture by the time M bovis was cultured from the milk samples.

The prevalence of M bovis–infected animals varied markedly among age cohorts (Table 2). The prevalence was highest in heifer calves 4 to 7 months old (33%), followed by cows 25 to 60 months old (18%). Interestingly, only 1 of 199 cattle 8 to 24 months old was confirmed to be infected with M bovis.

Table 2—

Age distribution for the M bovis–infected and uninfected cattle of Table 1.

 M bovis infection status 
Age (mo)PositiveNegativeNo. of cattle in age group
4–716 (33)33 (67)49
8–150 (0)61 (100)61
16–241 (1)137 (99)138
25–6034 (18)154 (82)188
61–782 (13)13 (87)15

Values represent the number (percentage) of animals within the given age group unless otherwise indicated.

See Table 1 for remainder of key.

Apparent sensitivity and specificity and positive and negative predictive values of antemortem bTB tests for the index herd

The apparent sensitivity and specificity were 86.8% and 92.7%, respectively, for the CFT alone and 80.4% and 96.5%, respectively, for the γIFN assay alone. The positive and negative predictive values were 61.3% and 98.1%, respectively, for the CFT alone and 75.9% and 97.3%, respectively, for the γIFN assay alone. When the results of the CFT and γIFN assay were collectively considered such that a test-positive animal was defined as an animal with a positive result on either the CFT or γIFN assay, the apparent sensitivity and specificity of the parallel test protocol was 96.1% and 91.7%, respectively, and the positive and negative predictive values were 61.3% and 99.4%, respectively.

Mycobacteria other than M bovis isolated from the index herd

Mycobacterium nonchromogenicum and Mycobacterium duvalii were isolated from the milk of 2 and 1 cows, respectively, that were confirmed to be infected with M bovis. Mycobacterium holsaticum was isolated from the milk of 1 M bovis–infected cow and 1 cow that was not confirmed to be infected with M bovis. Mycobacterium avium subsp hominissuis (n = 1 calf), Mycobacterium fortuitum (1), and M nonchromogenicum (1) were isolated from tissues of 3 calves that were not confirmed to be infected with M bovis.

Necropsy results for the cats and dog from the index herd

Of the 21 cats necropsied, 5 had lesions consistent with bTB, and 12 had M bovis cultured from tissues. The distribution of lesions and M bovis–culture positive tissues within the body of those cats has been described elsewhere.18 No evidence of bTB was found in the dog.

Results of bTB surveillance in wildlife

Eighty wild WTD were culled and necropsied by USDA Wildlife Services; 72 and 8 were culled within a 3.2- or 16-km radius of the index herd, respectively. None of the deer had gross or histologic lesions consistent with bTB, and M bovis was not isolated from any of the tissue specimens submitted for mycobacterial culture. Another 2,206 WTD harvested by hunters in Saginaw County during the fall hunting seasons of 2013, 2014, and 2015 were evaluated, and no evidence of bTB was found. Since bTB surveillance in wild WTD began in 1995, 12,100 deer from Saginaw and its contiguous (Arenac, Clinton, Genesee, Gratiot, Midland, Shiawassee, and Tuscola) counties have been examined, of which only 1 (a deer harvested by a hunter in Shiawassee County in 2007) was confirmed to be infected with M bovis.27

Fourteen raccoons, 5 opossums, and 1 coyote were trapped on the index herd property and necropsied. None of those animals had gross or histologic lesions consistent with bTB, and M bovis was not isolated from any of the tissue specimens submitted for mycobacterial culture.

Tracing of cattle and cattle products to and from the index herd

During the 5 years prior to identification of bTB in the index cow, no cattle were introduced into the index herd from the bTB-endemic area of the lower peninsula of Michigan. In fact, the only outside cattle introduced into the index herd during that period were mature bulls for breeding purposes. All bulls were obtained from 2 nearby herds, and 1 bull had been born into a third nearby herd. All 3 of those herds were screened for bTB, with no evidence of the disease found. Likewise, no evidence of bTB was found in any of the 37 cattle herds located within a 16-km radius of the index herd.

The only cattle that left the index herd were cull cows, which were sold for slaughter, and male calves < 1 week old, which were primarily sold to a producer in the neighboring Gratiot County, who had a calf-raising and feedlot operation (Gratiot operation) as well as friends and neighbors. Unpasteurized waste milk from the index herd was fed to calves retained as herd replacements and given away to other producers, friends, and neighbors to feed calves. The primary recipient of waste milk from the index herd was the Gratiot operation. Beginning in 2012, male calves left the index herd without being individually identified and therefore were not directly traceable.

At the Gratiot operation, unidentified calves from the index herd were mixed with calves from other herds, many of which were also unidentified. The owner of the Gratiot operation sorted all calves on arrival. Healthy calves were immediately resold to a livestock broker and removed from the operation, whereas small and poor-performing calves were maintained at his calf-raising facility and often fed unpasteurized waste milk from the index herd. Beginning in August 2013, the owner of the Gratiot operation applied his premise-specific radio-frequency identification tags to all unidentified commingled calves prior to their sale and removal from the operation. Healthy calves sold to the livestock broker were resold to a number of other calf-raising or backgrounding (where weaned calves are fed for a while before being resold to a feedlot) operations or feedlots.

Because of the lack of individual animal identification and insufficient records of animal movements, it was not possible to track or account for all cattle that left the index herd and Gratiot operation. Nevertheless, 3 M bovis–infected steers with direct links to the index herd were identified. One was found in a beef herd in Midland County (Midland herd), and the other 2 were identified in the Gratiot operation. Additionally, another M bovis–infected steer was detected in a feedlot in Arenac County (Arenac feedlot). That steer was purchased from the Gratiot operation and had consumed waste milk from the index herd prior to weaning.

During the epidemiological investigation for the index herd, routine slaughter surveillance in Michigan abattoirs identified 2 M bovis–infected steers, which were traced back to 2 feedlots in Huron County (Huron1 and Huron2 feedlots). Sixty-seven M bovis–infected cattle were subsequently identified in the Huron2 feedlot. Two additional M bovis–infected steers were detected during slaughter surveillance at a Wisconsin abattoir (Iowa steers). Those 2 steers were moved without authorization from Michigan to a backgrounding facility in Iowa, then through a livestock auction market in South Dakota to a feedlot in Nebraska before being sent to slaughter in Wisconsin. It could not be determined whether any of those 4 steers originated from the index herd, were fed waste milk from the index herd, or were exposed to another animal that was infected with the index herd strain of M bovis.

All cattle herds within a 5-km (3-mile) radius of the 5 Michigan bTB-affected herds (Gratiot operation, Midland herd, and Arenac, Huron1, and Huron2 feedlots) associated with the index herd outbreak were screened for bTB. To date, 107 herds containing 11,438 cattle have been tested in Arenac, Gratiot, Huron, Midland, and Saginaw Counties, and no additional bTB-affected operations have been identified.

WGS results

Eighty-one M bovis isolates from 73 animals underwent WGS during the outbreak investigation. Forty-eight of those isolates came from the index herd, and the other 33 isolates came from M bovis–infected cattle traced to other herds. Of the 48 isolates that were sequenced from the index herd, 28 were obtained from cows, 14 were obtained from calves, and 6 were obtained from cats. Two M bovis isolates were sequenced for each of 3 cows and 5 calves, and the 2 isolates sequenced from each of those animals were obtained from different body regions (head, thorax, abdomen, or milk). No SNP diversity was found within the 3 cows and 2 of the 5 calves. The SNP profiles for the remaining 3 calves differed slightly, which suggested that those animals had multiple sites of primary invasion and infection, implying that the source of M bovis for those calves was pooled milk from multiple cows that contained organisms with slightly different genotypes.

Of the 33 sequenced M bovis isolates that were obtained from cattle traced to herds other than the index herd, 26 were obtained from cattle in the Huron2 feedlot, 2 were obtained from steers from the Gratiot operation, 2 were obtained from the Iowa steers (identified as infected at the Wisconsin abattoir), and 1 each was obtained from steers traced back to the Midland herd, Arenac feedlot, and Huron1 feedlot. There was a low level of genetic diversity among the M bovis isolates associated with this outbreak. Results of WGS indicated that all trace-back premises except the Midland herd had cattle infected with 1 of the 2 major M bovis genotypes identified in the infected cattle of the index herd (Figure 2). The Midland herd had only 1 M bovis–infected animal, and the isolate sequenced from that animal was only 1 SNP different from the second most frequently identified genotype for the index herd. The WSG results for 6 isolates (obtained from 1 cow and 1 calf from the index herd and 4 steers from the Huron2 feedlot) had mixed calls at ≥ 2 loci, which suggested that those animals were infected with mixed populations (ie, multiple genotypes) of M bovis.

Figure 2—
Figure 2—

Graphical representation of the SNP differences as determined by WGS for Mycobacterium bovis isolates that were obtained from animals in Michigan. Each circle represents a specific M bovis genotype, and the size of the circle is proportional to the frequency with which that genotype has been identified. Numbers within shaded areas represent the number of isolates identified from a particular operation or herd with the same genotype if that number was > 1. The length of the line connecting 2 circles approximates how many SNPs differ between the 2 genotypes (ie, the shorter the line, the more similar the 2 genotypes are), and the length of 1 SNP is indicated in the illustration. Genotypes depicted above the horizontal dotted line were isolated from animals in the northeastern portion of the lower peninsula where M bovis is endemic in wild WTD (bTB-endemic area), and those depicted below the dotted line were isolated from animals outside the endemic area (bTB-free area) that were associated with the 2013 bTB outbreak investigation described in Figure 1. An M bovis strain isolated in 1994 from a wild WTD (1994 index deer) in the bTB-endemic area was the ancestral genotype for all genotypes sequenced from the animals associated with this outbreak and differed by only 9 SNPs from the most ancestral genotype of M bovis sequenced from cattle in the index herd. The hypothesized most recent common ancestor genotype (A) most likely originated in a WTD from the bTB-endemic area (no isolates with this genotype have been sequenced) and was only 2 SNPs different from the most ancestral genotype sequenced from the index herd. Only 2 genotypes within this lineage were isolated from cattle (1 from a cull dairy cow in Alpena County in 1993 [1993 dairy cow] and 1 from a beef cow in Alcona County in 2003 [2003 beef herd]) prior to the 2013 outbreak in the index herd. Both of those genotypes were 10 to 11 SNPs different from the most ancestral genotype sequenced from the index herd, which made them an unlikely source for the 2013 outbreak. See Figure 1 for remainder of key.

Citation: Journal of the American Veterinary Medical Association 251, 2; 10.2460/javma.251.2.206

An M bovis strain isolated in 1994 from a wild WTD (index deer) in the bTB-endemic area of Michigan was the ancestral genotype for all genotypes sequenced from the animals associated with this outbreak and differed by only 9 SNPs from the most ancestral genotype of M bovis sequenced from cattle in the index herd (Figure 2). The hypothesized most recent common ancestor genotype (A) most likely originated in a WTD from the bTB-endemic area (no isolates with this genotype have been sequenced) and was only 2 SNPs different from the most ancestral genotype sequenced from the index herd. Only 2 M bovis genotypes within this lineage have been isolated from cattle (1 from a dairy cow in Alpena County in 1993 and 1 from a beef cow in Alcona County in 2003) prior to the 2013 bTB outbreak associated with the index herd in Saginaw County. Both of those genotypes had divergent SNPs and were 10 to 11 SNPs different from the most ancestral genotype sequenced from the index herd, which made them an unlikely source for the 2013 outbreak. Several closely related M bovis isolates have been detected in wild WTD within the bTB-endemic area. Two of those isolates have a direct common ancestor genotype and were 3 SNPs different from the 1994 index deer and 6 SNPs different from the most ancestral genotype sequenced from the index herd. The WGS results for the M bovis strain isolated from an infected wild WTD harvested in Shiawassee County by a hunter in 2007 was only distantly related to the M bovis strains isolated from the cattle of the index herd and associated trace-back premises.

Discussion

The bTB outbreak described in the present report was unusual in several ways. The geographic locations of all bTB-affected herds involved in this outbreak were a considerable distance from the northeastern portion of the lower peninsula of Michigan where bTB is endemic in wildlife (bTB-endemic area). Also, the index herd of this investigation had a much higher prevalence of M bovis–infected animals (53/451 [11.8%]) than other bTB-affected herds identified in Michigan, and the distribution of infected animals varied markedly among age cohorts in an unusual manner. In other bTB-affected herds, the prevalence of M bovis–infected cattle increases linearly with age.28,29 For the index herd, the prevalence of M bovis–infected cattle was greatest for calves ≤ 7 months old (16/49 [33%]) followed by cows 25 to 60 months old (34/188 [18%]) and cows > 60 months old (2/15 [13%]). Finally, unpasteurized waste milk was identified as a likely and substantial source for M bovis transmission to cattle within the index herd and to other herds as well as to farm cats.

Whole-genome sequencing and SNP analysis facilitate assessment of the relatedness of M bovis isolates and can be useful in disease investigations. Whole-genome sequencing is much better at differentiating mycobacterial strains than spoligotyping and assessment of VNTR and restriction fragment length polymorphism patterns.30 Some M bovis isolates from Mexico and Michigan share the same spoligotype and VNTR pattern, likely owing to the historical trade of livestock between the United States and Mexico, and when bTB reemerged in Michigan in the late 1990s, USDA officials were concerned about the inability to differentiate between those strains. With the development of WGS, this is no longer a concern because M bovis isolates from Michigan differ by at least 50 SNPs from the common ancestor of M bovis isolates from Mexico.

For the bTB outbreak investigation described in the present report, results of WGS provided strong evidence that the movement of M bovis–infected cattle and contaminated unpasteurized waste milk from the index dairy herd was responsible for transmission of the disease to the affected beef operations. Most importantly, 3 of the 4 bTB-affected feedlots were linked to the index herd only on the basis of results of WGS and SNP analysis of M bovis isolates obtained from infected cattle because there was no record of movement of livestock or waste milk from the index herd to those operations. Because WGS can differentiate between M bovis strains and show how closely different strains are related, this technique has become an integral part of the USDA bTB eradication program.

Many factors can affect CFT and γIFN assay results, such as the age of animals being tested and cross-reactivity between M bovis and other mycobacteria.31,32 Mycobacteria other than M bovis were detected in milk samples from 5 cows and tissue specimens from 3 calves. All 5 cows tested positive to both the CFT and γIFN assay, but only 4 were confirmed to be infected with M bovis. All 3 calves tested positive on the CFT, and 2 also tested positive on the γIFN assay, but were not confirmed to be infected with M bovis. Although the number of cattle in the index herd infected with mycobacteria other than M bovis was low, results of this outbreak investigation suggested there may be cross-reactivity among mycobacterial species that can cause false-positive results on the CFT and γIFN assay.

The sensitivity of a diagnostic test is a measure of its ability to detect infected animals and is calculated as the proportion of truly infected animals that have a positive test result. The specificity of a diagnostic test is a measure of its ability to correctly identify noninfected animals and is calculated as the proportion of truly noninfected animals that have a negative test result. To measure the true sensitivity and specificity of a test, it is necessary to know the true infection status of each animal. It is widely known that some M bovis–infected animals do not have detectable bTB lesions. Therefore, use of PCR analysis or mycobacterial culture of tissue lesions as the standard to definitively determine bTB status will likely result in misclassification of some M bovis–infected animals as uninfected, and calculation of test sensitivity and specificity on the basis of such a standard (such as was done in the present investigation) should be considered the apparent rather than true sensitivity and specificity. When considered separately, the apparent sensitivity was 86.8% for the CFT and 80.4% for the γIFN assay; however, because it was possible that some M bovis–infected cattle were misclassified as uninfected, it is likely that the true sensitivity of both tests is higher. The apparent specificity was 92.7% for the CFT and 96.5% for the γIFN assay, and given that it was unlikely cattle not infected with M bovis would be misclassified as infected, the apparent specificity is probably a close approximation of the true specificity for both tests.

The positive predictive value of a diagnostic test is the probability that animals with positive results are truly infected, whereas the negative predictive value of a diagnostic test is the probability that animals with negative test results are truly uninfected. Predictive values are particularly sensitive to the prevalence of infection within the population being tested. As prevalence increases, the positive predictive value increases and the negative predictive value decreases, whereas when the prevalence decreases, the positive predictive value decreases and the negative predictive value increases. Therefore, if some M bovis–infected cattle were misclassified as uninfected (ie, prevalence was underestimated), the calculated positive predictive values would likewise be underestimated and the negative predictive values would be overestimated. Regardless, the predictive values reported for the CFT and γIFN assay for the index herd of this outbreak investigation should be interpreted with caution, particularly given that the apparent prevalence of M bovis–infected cattle in that herd was higher than that in other bTB-affected herds identified in Michigan.

Eradication of bTB is dependent on identifying and removing M bovis–infected animals; therefore, increasing the sensitivity of antemortem bTB testing protocols is crucial. For the index herd of the present investigation, apparent sensitivity was maximized (96.1%) when the CFT and γIFN assay results were interpreted in parallel, or collectively (ie, a test-positive animal was defined as an animal with a positive test result on either the CFT or γIFN assay). However, the apparent specificity when the CFT and γIFN assay results were interpreted in parallel (91.7%) was only slightly decreased from the specificity for each test when the results were interpreted separately. Theoretically, the increase in sensitivity from performing the CFT and γIFN assay in parallel will decrease the number of cattle with false-negative results and increase the number of diseased animals identified and removed, which will help decrease the spread of disease with a minimal increase in immediate costs. The tradeoff is that the decrease in specificity will result in an increase in number of animals with false-positive results that are unnecessarily removed from the population and increase the cost of the eradication program. However, in the case of the index herd, the decrease in specificity was small, and the negative predictive value when the CFT and γIFN assay results were interpreted in parallel was 99.4% (341/343), which indicated that the likelihood a test-negative animal was infected with M bovis was very small. Although the owner of the index herd of this investigation chose to depopulate the herd instead of remaining under quarantine in accordance with guidelines established by the USDA bTB eradication program,4 in another situation, use of the CFT and γIFN assay in parallel should maximize the number of infected animals identified and removed while providing reasonable assurance that animals with negative test results are truly not infected with M bovis.

The original source of M bovis for the index herd was not definitively determined during the present epidemiological investigation, but the WGS results indicated that the most probable source was the purchase of M bovis–infected cattle. The most ancestral M bovis genotype identified from infected cattle of the index herd differed by only 2 SNPs from M bovis genotypes isolated from animals in the bTB-endemic area of Michigan, although no direct link to previously identified M bovis isolates from cattle was found. Of the M bovis isolates currently indexed in the NVSL database, those that were most closely related to those sequenced in the present investigation were recovered from wild WTD harvested in the bTB-endemic area of Michigan between 1997 and 2001 and an opossum collected in 2013. Because none of the wildlife trapped or harvested on or around the property of the index herd had evidence of bTB, it is unlikely that local wildlife played a role in this outbreak. Moreover, given that the estimated home range of wild WTD in the northern portion of the lower peninsula of Michigan is only 3.1 to 3.4 km2 (1.9 to 2.1 miles2),21 it was also unlikely that M bovis–infected deer from the bTB-endemic area traveled approximately 160 km south and came into contact with the cattle of the index herd. In recent years, the only outside animals introduced into the index herd were a small number of bulls from nearby herds, all of which were screened without any evidence of bTB found. Consequently, the most likely source of M bovis for the index herd was infected cattle that were purchased from the bTB-endemic area of Michigan approximately 20 years earlier when it was actively expanding. Unfortunately, movement records for those cattle were not available, so we were unable to definitively determine whether that was indeed the source and can only speculate on the basis of the information available.

Multiple factors may have contributed to the unusual distribution of M bovis–infected cattle within the index herd. The high prevalence of M bovis–infected calves between 4 and 7 months old was likely the result of feeding unpasteurized waste milk to calves. Waste milk was typically obtained from multiple cows and pooled on a daily basis before being fed to calves. Given the fairly high prevalence of M bovis–infected cows in the lactating herd (36/203 [18%]), the probability that calves would ingest M bovis–contaminated milk sometime before being weaned was likewise fairly high.

Pasteurization has long been recognized as an effective method to control the spread of tuberculosis in milk and milk products.33 Recent advances in pasteurization technology now allow colostrum to be pasteurized without the destroying the vital immunoglobulins it contains.34 Pasteurization of colostrum decreases its pathogen load and provides a safer alternative than unpasteurized colostrum for feeding to neonatal calves. Although feeding unpasteurized milk to calves is a common practice, the present bTB outbreak investigation was the first in which unpasteurized waste milk from a bTB-affected herd was documented to have been shared with other herds. This was also the first investigation in which unpasteurized milk was implicated in the spread of bTB to cats and cattle in multiple herds.

In the United States, there has been a recent resurgence in the consumption of unpasteurized or raw milk by humans because of its perceived health benefits. This has resulted in efforts to legalize the sale of raw milk in some states, whereas in other states, individuals have used various means to bypass milk pasteurization laws to acquire raw milk for their own consumption. Those efforts have not abated despite numerous instances of human disease and deaths associated with the consumption of unpasteurized milk and milk products.35,36 Dairy farm families commonly consume unpasteurized milk from their own cows. In Michigan, public health officials are notified whenever a bTB-affected herd is identified, and those officials consult with and provide tuberculosis testing for anyone who might have been exposed to M bovis. Because of patient confidentiality laws, information regarding possible M bovis infections in humans associated with this investigation was not available.

In the present bTB outbreak investigation, M bovis was cultured from the milk of 2 of the 19 cows from which milk samples were obtained. One of those cows was positive on both the CFT and γIFN assay, whereas the other was positive on the γIFN assay only. Cows with negative results on bTB skin tests can shed M bovis in milk.37 Therefore, although screening cattle for bTB prior to consumption of raw milk produced by those cows is recommended, a negative test result does not guarantee that the milk will be free of M bovis.

Other farm management practices that likely contributed to bTB transmission within the index herd included overcrowding of cattle, frequent movement and regrouping of calves, lack of separation among cattle of different ages, and inadequate colostrum consumption by neonatal calves (colostrum was mixed with waste milk and fed to all calves regardless of age), which impaired the acquisition of passive immunity by those calves.34

The lack of individual-animal identification and livestock movement records greatly impeded this bTB outbreak investigation and was a violation of state and federal regulations. There were also no records regarding the distribution of waste milk from the index herd. This resulted in the tracking and testing of thousands of cattle that may not have had any association with cattle or waste milk from the index herd at a considerable cost. Although all livestock herds that were epidemiologically linked to the index herd or located within the USDA-mandated radius (16 km for the index herd and Midland herd and 4.8 km [3 miles] for the 4 feedlots) of all bTB-affected operations were screened for the disease, it is possible some bTB-affected herds or M bovis–infected animals were not identified by the methods used. An education campaign directed at livestock owners and dealers by the Michigan Department of Agriculture and Rural Development has greatly increased regulatory compliance.

In summary, the bTB outbreak investigation described in this report was unique in that results of WGS were used to link M bovis–infected cattle with the index herd that were not otherwise identified by tracing the movement of animals or animal products to and from the index herd. This allowed for a more comprehensive investigation and suggested that WGS should be used in all disease investigations because it is better at differentiating bacterial strains than spoligotyping or analyzing VNTR and restriction fragment length polymorphism patterns. In this investigation, the apparent sensitivity and specificity were 86.8% and 92.7%, respectively, for the CFT and 80.4% and 96.5%, respectively, for the γIFN assay when the results for those tests were interpreted separately but were 96.1% and 91.7% when the CFT and γIFN assay test results were interpreted in parallel. Those findings suggested that the CFT and γIFN assay should be performed in parallel to maximize the sensitivity of antemortem bTB testing for the identification of M bovis–infected cattle. Also unique to this investigation was the fact that unpasteurized waste milk was identified as an important risk factor for M bovis transmission to cattle within the index herd and to other herds as well as to farm cats.

Acknowledgments

No third-party funding or support was received in connection with this study or the writing or publication of the manuscript. The authors declare that there were no conflicts of interest.

The authors thank Drs. Kathleen Orloski, Tolani Francisco, and Gary Stevens for their expertise with this disease investigation; Dr. Joe Woltanski for technical assistance; Dr. Vicki Chickering for coordinating the farm investigations; Phyllis Rayca and Melinda Cosgrove for assistance with manuscript preparation; and Patrick Ryan for assistance with data collection.

ABBREVIATIONS

AC

Allele count

bTB

Bovine tuberculosis

CFT

Caudal fold test

DCPAH

Diagnostic Center for Population and Animal Health

γIFN

γ-Interferon

NVSL

National Veterinary Services Laboratories

SNP

Single nucleotide polymorphism

VNTR

Variable number tandem repeat

WGS

Whole-genome sequencing

WTD

White-tailed deer

Footnotes

a.

Bovigam, Thermo Fisher Scientific, Waltham, Mass.

b.

MGIT, Becton Dickinson, Sparks, Md.

c.

Nextera XT Library Sample Preparation Kitd and MiSeq, Illumina, San Diego, Calif.

d.

USDA NVSL Bacterial SNP Pipeline, USDA Veterinary Services, Washington, DC. Available at: github.com/USDA-VS. Accessed Jan 31, 2017.

e.

GATK and HaplotypeCaller, version 3.2, Broad Institute, Cambridge, Mass.

References

  • 1. Schmitt SM, Fitzgerald SD, Cooley TM, et al. Bovine tuberculosis in free-ranging white-tailed deer in Michigan. J Wildl Dis 1997; 33: 749758.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 2. Okafor CC, Grooms DL, Bruning-Fann CS, et al. Descriptive epidemiology of bovine tuberculosis in Michigan (1975–2010): lessons learned. Vet Med Int 2011; 2011: 874924.

    • Search Google Scholar
    • Export Citation
  • 3. Kaneene JB, VanderKlok M, Bruning-Fann CS, et al. Prevalence of Mycobacterium bovis infection in cervids on privately owned ranches. J Am Vet Med Assoc 2002; 220: 656659.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 4. USDA APHIS. Bovine tuberculosis eradication uniform methods and rules, effective January 1, 2005. Available at: www.aphis.usda.gov/animal_health/animal_diseases/tuberculosis/downloads/tb-umr.pdf. Accessed Mar 31, 2017.

  • 5. USDA Food Safety and Inspection Service. Guideline Number 4. Inspection of tuberculin reactors. Available at: www.fsis.usda.gov/wps/wcm/connect/9b50790b-3875-4a76-a565-8fc7f374043e/TB_Guideline_4_88.pdf?MOD=AJPERES. Accessed Mar 31, 2017.

    • Crossref
    • Export Citation
  • 6. Miller JM, Jenny AL, Payeur JB. Polymerase chain reaction detection of Mycobacterium tuberculosis complex and Mycobacterium avium organisms in formalin-fixed tissues from culture-negative ruminants. Vet Microbiol 2002; 87: 1523.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 7. Robbe-Austerman S, Bravo DM, Harris B. Comparison of the MGIT 960, BACTEC 460 TB and solid media for isolation of Mycobacterium bovis in United States veterinary specimens. BMC Vet Res 2013; 9: 74.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 8. Bradner L, Robbe-Austerman S, Beitz DC, et al. Chemical decontamination with N-acetyl-l-cysteine–sodium hydroxide improves recovery of viable Mycobacterium avium subsp paratuberculosis organisms from cultured milk. J Clin Microbiol 2013; 51: 21392146.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 9. Higgins J, Camp P, Farrell D, et al. Identification of Mycobacterium spp of veterinary importance using rpoB gene sequencing. BMC Vet Res 2011; 7: 77.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 10. Li H, Durbin R. Fast and accurate short read alignment with Burrows-Wheeler transform. Bioinformatics 2009; 25: 17541760.

  • 11. Li H, Handsaker B, Wysoker A, et al. The sequence alignment/map format and SAMtools. Bioinformatics 2009; 25: 20782079.

  • 12. DePristo MA, Banks E, Poplin R, et al. A framework for variation discovery and genotyping using next-generation DNA sequencing data. Nat Genet 2011; 43: 491498.

    • Search Google Scholar
    • Export Citation
  • 13. Van der Auwera GA, Carneiro MO, Hartl C, et al. From FastQ data to high confidence variant calls: the Genome Analysis Toolkit best practices pipeline. Curr Protoc Bioinformatics 2013; 11: 11.10.111.10.33.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 14. McKenna A, Hanna M, Banks E, et al. The Genome Analysis Toolkit: a MapReduce framework for analyzing next-generation DNA sequencing data. Genome Res 2010; 20: 12971303.

    • Search Google Scholar
    • Export Citation
  • 15. International Union of Pure and Applied Chemistry and International Union of Biochemistry and Molecular Biology Joint Commission on Biochemical Nomenclature and Nomenclature Commission of IUBMB. In: Liebecq C, ed. Biochemical nomenaclature: and related documents: a compendium. 2nd ed. London: Portland Press Ltd, 1992; 122126.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 16. Stamatakis A. RAxML version 8: a tool for phylogenetic analysis and post-analysis of large phylogenies. Bioinformatics 2014; 30: 13121313.

  • 17. Robinson JT, Thorvaldsdóttir H, Winckler W, et al. Integrative genomics viewer. Nat Biotechnol 2011; 29: 2426.

  • 18. Fitzgerald SD, Hollinger C, Mullaney TP, et al. Herd outbreak of bovine tuberculosis illustrates that route of infection correlates with anatomic distribution of lesions in cattle and cats. J Vet Diagn Invest 2016; 28: 129132.

    • Search Google Scholar
    • Export Citation
  • 19. Memorandum of Understanding Between Michigan Department of Agriculture and Rural Development and the Michigan Department of Natural Resources and the United States Department of Agriculture. Available at: www.michigan.gov/documents/emergingdiseases/MOU_TB_MDARD_MDNR_USDA_signed_051916_526238_7.pdf. Accessed Mar 31, 2017.

    • Crossref
    • Export Citation
  • 20. USDA APHIS. Guidelines for surveillance of bovine tuberculosis in wildlife. June 2011. Available at: www.aphis.usda.gov/animal_health/animal_diseases/tuberculosis/downloads/wildlife_tb_surv_manual.pdf. Accessed Mar 31, 2017.

    • Crossref
    • Export Citation
  • 21. Michigan Department of Natural Resources. Michigan deer management plan. Appendix D: a review of deer management in Michigan. Available at: www.michigan.gov/dnr/0,4570,7-153-10363_41153—,00.html. Accessed Mar 3, 2016.

    • Crossref
    • Export Citation
  • 22. O'Brien DJ, Fitzgerald SD, Lyon TJ, et al. Tuberculous lesions in free-ranging white-tailed deer in Michigan. J Wildl Dis 2001; 37: 608613.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 23. O'Brien DJ, Schmitt SM, Fierke JS, et al. Epidemiology of Mycobacterium bovis in free-ranging white-tailed deer, Michigan, USA, 1995–2000. Prev Vet Med 2002; 54: 4763.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 24. Bruning-Fann CS, Schmitt SM, Fitzgerald SD, et al. Mycobacterium bovis in coyotes from Michigan. J Wildl Dis 1998; 34: 632636.

  • 25. Bruning-Fann CS, Schmitt SM, Fitzgerald SD, et al. Bovine tuberculosis in free-ranging carnivores from Michigan. J Wildl Dis 2001; 37: 5864.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 26. Walter WD, Fischer JW, Anderson CW. Surveillance and movements of Virginia opossum (Didelphis virginiana) in the bovine tuberculosis region of Michigan. Epidemiol Infect 2013; 141: 14981508.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 27. O'Brien DJ, Schmitt SM, Fitzgerald SD, et al. Management of bovine tuberculosis in Michigan wildlife: current status and near term prospects. Vet Microbiol 2011; 151: 179187.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 28. Brooks-Pollock E, Conlan AJ, Mitchell AP, et al. Age-dependent patterns of bovine tuberculosis in cattle. Vet Res 2013; 44: 97.

  • 29. Francisco TI, Orloski KA, Roberts NJ. Investigation of a Mycobacterium bovis outbreak in cattle at a Colorado dairy in 2010. J Am Vet Med Assoc 2014; 244: 805812.

    • Search Google Scholar
    • Export Citation
  • 30. Roetzer A, Diel R, Kohl TA, et al. Whole genome sequencing versus traditional genotyping for investigation of a Mycobacterium tuberculosis outbreak: a longitudinal molecular epidemiological study. PLoS Med 2013; 10:e1001387.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 31. de la Rua-Domenech R, Goodchild AT, Vordermeier HM, et al. Ante mortem diagnosis of tuberculosis in cattle: a review of the tuberculin tests, γ-interferon assay and other ancillary diagnostic techniques. Res Vet Sci 2006; 81: 190210.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 32. Cagiola M, Feliziani F, Severi G, et al. Analysis of possible factors affecting the specificity of the gamma interferon test in tuberculosis-free cattle herds. Clin Diagn Lab Immunol 2004; 11: 952956.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 33. Holsinger VH, Rajkowski KT, Stabel JR. Milk pasteurisation and safety: a brief history and update. Rev Sci Tech 1997; 16: 441451.

  • 34. Hurley WL, Theil PK. Perspectives on immunoglobulins in colostrum and milk. Nutrients 2011; 3: 442474.

  • 35. Claeys WL, Cardoen S, Daube G, et al. Raw or heated cow milk consumption: review of risks and benefits. Food Control 2013; 31: 251262.

  • 36. Langer AJ, Ayers T, Grass J, et al. Nonpasteurized dairy products, disease outbreaks, and state laws—United States, 1993–2006. Emerg Infect Dis 2012; 18: 385391.

    • Search Google Scholar
    • Export Citation
  • 37. Zarden CF, Marassi CD, Figueiredo EE, et al. Mycobacterium bovis detection from milk of negative skin test cows. Vet Rec 2013; 172: 130.

All Time Past Year Past 30 Days
Abstract Views 142 0 0
Full Text Views 685 597 96
PDF Downloads 172 80 4
Advertisement