Diagnosing Lymphoproliferative Disease Virus in Wild Turkeys (Meleagris gallopavo)
The Edna Bailey Sussman Foundation Internship Report
December 2014
Katrina Alger, MS Student
State University of New York College of Environmental Science and Forestry
Department of Environmental and Forest Biology
Background
Wild turkeys are an important species on the North American landscape, both for the role they play in forest ecosystems, and their value as a game bird. One hundredyears ago, turkeys were nearly extirpated from many states due to habitat loss and overhunting (Tapley et al., 2005). Recovery efforts based on forest restoration and flock relocation were enormously successful, resulting in rebounding populations throughout much of the eastern half of the United States (Tapley et al., 2005). Careful management of this species continues to be a high priority for many states in order to prevent such a crash from recurring.
Over the past decade many states in the Northeast, including New York, have noticed a steady fall in annual harvest rates – the primary metric used by many natural resource agencies to estimate population size. This has lead wildlife managers to worry thatwild turkey numbers are again in decline. Coinciding with this downward trend was the discovery, in 2009, of lymphoproliferative disease virus (LPDV) in wild turkeys along the east coast (Allison et al., 2014). This virus was previously thought to only infect domestic turkeys where it was characterized by a rapid clinical course leading to flock mortality as high as 25 percent (Biggs et al., 1978).
The discovery of LPDV in wild turkeys has prompted many questions regarding its potential population impacts non-domesticated birds. Recent studies have indicated that the prevalence is likely quite high (exceeding 50% in many states) and that many birds infected with LPDV are asymptomatic (Thomas, 2013; Allison et al., 2014). The lack of overt disease signs combined with the fact that the virus elicits a poor antibody response in the host and no one has successfully cultured the virus in the lab (Patel and Shilleto, 1987), makes LPDV particularly difficult to diagnose in living birds. Definitive diagnoses have only been made post-mortem, using polymerase chain reaction (PCR) to detect the virus in bone marrow or other tissues (Allison et al., 2014). Lack of a reliable blood test has hindered research on important questions regarding LPDV, such as how the virus is being transmitted in wild birds, what triggers the onset of disease in birds asymptomatically infected with the virus, and whether or not it is contributing to the recently observed decline in turkey populations.
Internship Objectives
My internship with the Animal Health Diagnostic Center (AHDC) at Cornell University focused on:
1)Developing a reliable test for LPDV using whole blood of live, banded birds, giving us the ability to track those birds and monitor their fate
2)Gaining experience in molecular diagnostic techniques and wildlife pathology at a large veterinary diagnostic and research facility
Methods
Whole blood was evaluated as a potential diagnostic test for LPDV relative to buffy coat and bone marrow, which were chosen as the “gold standards” for diagnosing LPDV based on previous studies confirming the presence of the virus in these two tissues. During the winter of 2014, blood was collected in collaboration with the New York Department of Environmental Conservation (NYSDEC) as they banded wild turkeys throughout the state as part of a four-year hen survival study (Figure 1).
One hundred-thirty seven samples were collected using EDTA tubes to prevent coagulation. A small portion (25 µl) was reserved for testing and the remainder was centrifuged for 7 minutes at 1500 x g to separate the white blood cells (buffy coat). In addition to these samples, whole blood and bone marrow from 32 hunter-killed birds were provided by Dr. Gavin Hitchener, a veterinary pathologist at the AHDC.
Extraction of DNA was performed on all tissues using DNeasy 96 Blood and Tissue Kits (Qiagen, Valencia, California) according to the manufacturer’s guidelines and PCR was performed using primers designed by Allison et al. (2014). This existing PCR assay was optimized for use with blood by modifying the concentration of Magnesium Chloride (MgCl2) and the annealing temperature for the reactions. Confirmation of the correct identity (LPDV) of the resulting PCR product was conducted by DNA sequencing for a subset of samples (n=108), representing each capture site.
Results were considered either true positives or true negatives when they tested positive or negative by the gold standard method (buffy coat or bone marrow). False positives and negatives were samples that tested positive or negative by whole blood, but had the opposite result using the gold standard. True prevalence was calculated by dividing the number of positive buffy coat or bone marrow samples (true positives) by the total sample size; apparent prevalence was calculated by dividing the number of positive whole blood samples by the total sample size. Cohen’s kappa, a statistical measure of agreement for categorical data, was used to determine the level of agreement between whole blood and buffy coat or whole blood and bone marrow.
Results and Discussion
For 137 paired samples of whole blood and buffy coat, apparent prevalence was 82% (112/137) and true prevalence was 84% (115/137) (Table 1). Sensitivity for whole blood compared to buffy coat was 97% and the specificity 95%. Cohen’s kappa was significant (κ = 0.90, Z = 6.53, p < 0.00001)indicating almost perfect agreement between the two sample types. For the 32 paired samples of whole blood and bone marrow, apparent prevalence was 75% (24/32) with a true prevalence of 72% (23/32). Sensitivity of whole blood versus bone marrow was 100% and specificity was 89%.
Evaluating the accuracy of a new diagnostic method also entails the calculation of the test’s positive (PPV) and negative predictive values (NPV). These represent the accuracy of the test as represented by the proportion of positive or negative test results that reflect either true positives or true negatives in the population. Unlike sensitivity and specificity, the positive and negative predictive values of a test depend fluctuate depending upon the true prevalence of infection in a population (Friis and Sellers, 2014). Prior studies estimate the prevalence of LPDV in New York to be approximately 45% (Thomas, 2013), which is much lower than the prevalence in our sample. Therefore, positive and negative predictive values were calculated and graphed for a range of prevalence values, based on the sensitivity and specificity of whole blood versus buffy coat (Figure 2) and bone marrow (Figure 3). Our data indicate that whole blood will be an accurate diagnostic over 90% of the time, even in populations with prevalence that differ significantly from our study population (PPV and NPV ≥ 90% in populations with prevalence between 49-80%.
Table 1. Evaluation of whole blood as a diagnostic test against two different gold standards (buffy coat and bone marrow). Level of significance for Cohen’s kappa α= 0.05.
Whole Blood vs. Buffy Coat / Whole Blood vs. Bone MarrowResult / 95% CI (upper) / 95% CI (lower) / Result / 95% CI (upper) / 95% CI (lower)
Apparent prevalence / 82% / 75% / 88% / 75% / 57% / 89%
True prevalence / 84% / 77% / 90% / 72% / 53% / 86%
Sensitivity / 97% / 93% / 99% / 100% / 79% / 100%
Specificity / 95% / 77% / 100% / 89% / 52% / 100%
PPV* / 99% / 95% / 100% / 96% / 79% / 100%
NPV** / 88% / 68% / 97% / 100% / 52% / 100%
Cohen's kappa / 0.9 / 0.79 / 0.99 / 0.92 / 0.77 / 1.07
P-value / <0.0001 / <0.0001
*Positive Predictive Value **Negative Predictive Value
These results suggest that PCR of whole blood provides detection of LPDV on par with both buffy coat and bone marrow. This is valuable information because research on LPDV is hampered by the lack of a simple and reliable test for infection in living birds. While the virus has been detected in white blood cells (buffy coat) and collection of this sample can be done non-lethally, it also requires careful handling in the field to prevent clotting or freezing, and additional processing steps in the lab, making it a more cumbersome sample to work with and increasing the desirability for a whole blood alternative. There was some disagreement between different tissues, but this was statistically insignificant. The small degree of disagreement observed in both comparisons could be due to poor sample quality or DNA degradation from long storage time. As methods are refined, this may become less common.
Results of this study were presented in a poster (acknowledging the Edna Bailey Sussman Foundation) at the annual meeting of The Wildlife Society in Pittsburgh this October. They are now being prepared for publication in the peer reviewed literature with a target of January for submission. Future research will involve integrating these results with band returns during the fall and spring hunting seasons to gather baseline data on the longevity of birds infected with LPDV and assess any potential changes in physical condition that occurred over the intervening year. The DNA sequencing data will be analyzed for variation in viral genotype at the regional, township, or flock level, to investigate epidemiological linkages. In addition, tissue samples from hunter-harvested birds from 2012-2014 will be tested and combined with the data from this study to investigate potential relationships between LPDV infection and land use, in an effort to construct a risk-surface map for the virus in New York State.
Acknowledgments
I thank the Edna Bailey Sussman Foundation for the financial support that made possible my internship at the AHDC. In addition to learning valuable molecular diagnostic techniques that I can continue utilizing for my Masters research, I gained insights and experience in wildlife disease diagnostics and management. I also extend thanks to Dr. Elizabeth Bunting, for sponsoring my internship at the AHDC, and to Dr. Krysten Schuler and Dr. Christopher Whipps for their indispensable guidance on this project. Finally, I am deeply grateful to all of the NYSDEC game bird field crews who collected samples for me, and to Michael Schiavone who helped coordinate their efforts.
Literature Cited
Allison, A. B., Kevin Keel, M., Philips, J. E., Cartoceti, A. N., et al. (2014). Avian oncogenesis induced by lymphoproliferative disease virus: A neglected or emerging retroviral pathogen? Virology, 450-451, 2–12. doi:10.1016/j.virol.2013.11.037
Biggs, P. M., McDougall, J. S., Frazier, J. A, & Milne, B. S. (1978). Lymphoproliferative disease of turkeys 1. Clinical aspects. Avian Pathology, 7(1), 131–9. doi:10.1080/03079457808418265
Friis, R., & Sellers, T. (2014). Epidemiology for Public Health Practice (Fifth Edit.). Burlington, MA: Jones & Bartlett Learning.
Patel, J. R., & Shilleto, R. W. (1987a). Detection of lymphoproliferative disease virus by an enzyme-linked immunosorbent assay. Epidemiology and Infection, 99(3), 711–22
Tapley, J. L., Abernethy, R. K., Kennamer, J. E., & Carolina, N. (2005). Status and Distribution of the Wild Turkey in 2004. In C. A. Stewart & V. Frawley (Eds.), Wild Turkey Management: Accomplishments, Strategies, and Opportunities - Proceedings of the Ninth National Wild Turkey Symposium. Grand Rapids, Michigan: Michigan Department of Natural Resources.
Thomas, J. M. (2013). Molecular Surveillance for Lymphoproliferative Disease Virus in Hunter Killed Wild Turkeys (Meleagris gallopavo) from the Eastern United States. University of Georgia.
1