Item No. 1 of 1

INVESTIGATOR: Mitchell, R. M.; Schukken, Y. H.; Medley, G. F.

PERFORMING INSTITUTION:
CORNELL UNIVERSITY
ITHACA, NEW YORK 14853

MODELING TRANSMISSION DYNAMICS OF MYCOBACTERIUM AVIUM SUBSP. PARATUBERCULOSIS TO IMPROVE CONTROL STRATEGIES ON COMMERCIAL U.S. DAIRY FARMS

NON-TECHNICAL SUMMARY: Mycobacterium avium subspecies paratuberculosis (MAP) is a chronic infection of cows and sheep which is common to most dairy farms in the United States (68% in the most recent national survey). Animals usually do not test positive for MAP until they are adults, even though we believe the majority of animals are infected at a very young age. Clinical disease causes progressive diarrhea, wasting and decrease in milk production in dairy cattle. Understanding MAP transmission cycles allows us to better design control strategies which are cost effective for farmers. In this project, a seven-year longitudinal dataset of samples from three northeastern US dairy farms will be analyzed to better understand how MAP is being transmitted between animals on the farms. We will look for associations between exposure to highly infectious adult animals and becoming infected: calves from infected dams, clusters of infected animals that were born on the same date, animals exposed to highly infectious animals as adults. We will test whether animals are commonly infected with more than one MAP strain (simulataneously, or throughout life) by testing composite samples from individual animals via dilution. This method will also be used to determine MAP diversity in positive environmental samples. Currently we assume that most farms have one strain of MAP on the farm, or that there is no competition between MAP strains. However, this project will be the first to examine long-term dynamics of MAP on farms with multiple strains present at the strain level. Based on the findings from the longitudinal dataset, we will construct mathematical models of MAP transmission that allow us to trace individual animals and assign duration of infection along specific distributions. Findings from this work will be disseminated in both scientific papers and meetings. These findings will aid veterinarians in advising farmers on best management practices to decrease clinical disease on US dairy farms.

OBJECTIVES: The overall goal of this project is to develop a mathematical model using precise longitudinal field and molecular typing data that will explain endemic persistence of MAP within dairy herds and allows evaluation of cost-effective MAP control strategies. To reach this long term goal, the following three specific objectives are defined: (1) Develop an expanded state transition model, accommodating new insights in infection transmission and latency to analyze a longitudinal dataset gathered by the Regional Dairy Quality Management Alliance and Johne's Disease Integrated Project. (2) Accurately estimate infection transmission parameters using molecular epidemiology and longitudinal field data. (3) Develop a stochastic, individual-based, model of infection transmission of MAP. This model is expected to more accurately predict infection endemicity and is suitable for development of MAP control programs. Findings from this work will be disseminated both in scientific literature and by short courses. Target date for completing analysis of MAP transmission at birth is June 2012. Target date for completing work on potential adult-to-adult transmission is early 2013, and final paper on MAP endemicity is June 2013. A short course on use of mathematical models for bovine veterinarians will be developed for Fall 2012 and a semester course for veterinary students will be proposed for Spring 2013.

APPROACH: Objective I: We will analyze whether animals are at higher risk of intermittent shedding surrounding stress events (calving, mastitis and lameness). Each sampling point will represent an independent event, and we will correct for repeated measurements of the same cow at multiple sampling points. We will analyze two subgroups of the population for association between adult exposure to shedding animals and infection status: (A) animals that were purchased as adults and (B) low shedding animals (fecal-culture or tissue-culture positive). Animals that culture positive in tissue or fecal samples will be strain typed using a multi-locus short sequence repeat typing scheme. For animals that are consistently low or non-shedding (B), time in the same pen with a high shedder will be used as a predictor for MAP strain or positive infection status. Other parameters will be included if significant in preliminary univariate analyses (lactation number, birth cohort of high shedder, stage of lactation, etc). A subset of fecal culture positive animals will be analyzed using single genome analysis (SGA) to assess MAP co-infections. This is a novel approach for addressing within-host diversity in MAP, but has been used successfully in viral quasi-species analysis including in HIV to increase sensitivity of detection of low prevalence strains within a host. Methods for this technique are identical to the normal extraction and sequencing protocol with one exception: Prior to PCR, bacterial DNA will be diluted until only 1/3 of wells have template copies and produce an amplicon. With this dilution, each well which produced an amplified product will have on average only one genomic template copy of DNA. Potential highly susceptible animals will be first priority for SGA. Objective II: Clusters of calves born at periods of high risk of infection will be identified using the Bernoulli purely temporal algorithm in SatScan. We will genotype the MAP isolates recovered from three sets of samples: (A) MAP-positive dam-daughter pairs, (B) clusters of MAP-positive calves, and (C) environmental samples at sampling times surrounding cluster outbreaks. Objective III: We will construct event-based and individual-based stochastic models to reflect new compartments and parameter estimates. Individual-based models will have abrupt transition between age classes. These models will allow us to eventually model infection and intervention strategies over identical population dynamics to view impact of control strategies on identical populations. Model framework will include age-dependent contact structure using a WAIFW (Who Acquires Infection From Whom) matrix as well as stage-of-lactation dependent contact structure for adults.

PROGRESS: 2011/09 TO 2014/08
Target Audience: During this three month period, data was presented from this project at the National Institute for Mathematical Biology working group on MAP at the University of Tennessee (Dec 2013). As well, a presentation based on work done with mentee Nelly Marquetoux was given to the Agricultural University of Iceland (Dec 2013). Changes/Problems: There have been no additional changes to project design or timeline in this period. What opportunities for training and professional development has the project provided? Through this period I had the opportunity to work with the NiMBioS working group mentors, including Dr. Ynte Schukken (Cornell University), Dr. Yoram Louzoun (Bar Ilan University) and Dr. Ad Koets (Utrecht). How have the results been disseminated to communities of interest? Nothing Reported What do you plan to do during the next reporting period to accomplish the goals? Nothing Reported

IMPACT: 2011/09 TO 2014/08
What was accomplished under these goals? Impact statement: This project sought to evaluate the relationship between our assumptions of transmission of this pathogen and the real data we can gather from field studies like thye Regional Dairy Quality Management Alliance study on Mycobacterium avium subspecies paratuberculosis (MAP). This post-doctoral project capitalized on and augmented available data to analyze new questions and increase the value of the currently available dataset. Through this work, data from this project became available to multiple additional research groups, and the knowledge gained from this project was also involved in generation of several additional projects among collaborators. We understand better the patterns of MAP progression through this project, including the existence of a population of infected animals which do not necessarily progress to clinical disease. This information aids us in understanding not only what happens within farms but also within infected animals, allowing development of control strategies more appropriate for individual animals in a population. ---------------------------- (1) Develop an expanded state transition model, accommodating new insights in infection transmission and latency to analyze a longitudinal dataset gathered by the Regional Dairy Quality Management Alliance and Johne's Disease Integrated Project. The expanded model developed under objective 2 incorporating direct and indirect transmission pathways evaluated using the RDQMA data and molecular data generated through this project was submitted to Epidemics. A key outcome of this work was to affirm the contribution of direct transmission to MAP transmission within dairy environments with close contact between animals in a barn environment. (2) Accurately estimate infection transmission parameters using molecular epidemiology and longitudinal field data. The longitudinal data analysis to assess objective 2 continued in collaboration with Dr. Yoram Louzoun and Ynte Schukken over this three month period. Through working with the National Institute for Mathematical and Biological Synthesis (NimBIoS), we have expended from the three-farm dataset available through the initial project proposal to access data from more than 20 farms on which animals were longitudinally tested. A draft of results will be presented at the NimBioS within host MAP modeling meeting in June 2014 and submitted for publication. The evaluation of the impact of stress events including calving on MAP progression under objective 2 was submitted for review. A key outcome of this work was the finding that stress events surrounding calving did not substantially impact the probability of initiation of shedding or immune changes in the course of a lactation. A second key outcome indicated that younger animals are not at lower risk of becoming MAP shedders during their first lactations, which could lead to a change in testing strategy, where often animals are not tested for MAP during a first lactation due to perceived lower risk of becoming infectious.

PUBLICATIONS (not previously reported): 2011/09 TO 2014/08
1. Type: Journal Articles Status: Submitted Year Published: 2014 Citation: HM Mitchell, RM Mitchell, AK Pradhan, T Fyock, JL Khol, RH Whitlock, YH Schukken. Association of lactation number, lactation stage and periparturient stress with Mycobacterium avium subsp. paratuberculosis serologic and shedding status
2. Type: Journal Articles Status: Submitted Year Published: 2014 Citation: RM Mitchell,RH Whitlock, YT Grohn, YH Schukken. Back to the real world: Connecting models with data. Preventive Veterinary Medicine.

PROGRESS: 2012/09/01 TO 2013/08/31
Target Audience: During this time period, publications from Dr. Mitchell's work appeared in both Preventive Veterinary Medicine and Applied Environmental Microbiology journals. As well, I gave presentation at the National University of Mexico (Sept 2012), Cambridge University (November 2012), University of Iceland (March 2013), and two presentations on this work were presented at both the American Association for Bovine Practitioners and Epidemics conferences. The audience for this work was therefore a combination of scientists, veterinarian researchers and practitioners. Specifically both the community of field veterinarians and veterinary scientists have access to this information. Changes/Problems: There has been a change to an individual-based teaching strategy rather than a classroom teaching approach. Because of the valuable opportunity to develop research skills working with Dr. Medley, Dr. Mitchell's stay at Warwick University was extended from 3 to 6 months. To foster collaboration and enhance data analysis quality, Dr. Mitchell also spent one month at Bar Ilan University working with Dr. Yoram Bar and his student Noa Slater on data analysis and model fitting. The research schedule was extended to December 2013. What opportunities for training and professional development has the project provided? Dr Mitchell had the opportunity to work with four mentors (Dr Graham Medley: Warwick University, Dr. Ynte Schukken: Cornell University, Dr. Yoram Louzoun: Bar Ilan University, Dr. Cord Heuer: Massey University) over the 2012-2013 post-doctoral period as well as attend three short courses on mathematical modeling and statistics and participate in a mathematical modeling working group. Through her mentors, Dr. Mitchell has gained expertise in statistical analyses, individual-based model development, data analysis and maximum-likelihood based model fitting, and MAP transmission dynamics in pastoral environments. During the course of research, five students worked with Dr. Mitchell: Dr. Nelly Marquetoux, a PhD student at Massey University, Isabelle Louge and Casey Cazer, veterinary and undergraduate students at Cornell University, Noa Slater, a PhD candidate at Bar Ilan University, and Dr Hilda Mitchell, a veterinarian writing a diplomarbiet through the University of Vienna. How have the results been disseminated to communities of interest? Results have been disseminated to communities of interest via presentations and paper. Dr. Mitchell has given a series of invited talks to veterinary and mathematically focused research groups including the University of Tennessee NIMBioS group, Cambridge University, the National University of Mexico and the Agricultural University of Iceland. Four publications have been submitted over the course of the year prior to September 2013 and four more publications are in preparation. The directly field-applicable data from this project was presented at the major bovine veterinary conference in the US, and the research-focused data was submitted to a conference focused on mathematical modeling. What do you plan to do during the next reporting period to accomplish the goals? The papers in draft should be completed in the next reporting period. All projects are in draft, with submission dates targeted in early 2014.

IMPACT: 2012/09/01 TO 2013/08/31
What was accomplished under these goals? Goals 1 and 2: An expanded state-transition model was developed and tested using the RDQMA dataset. This paper was in draft at the end of Sept 2012, and an abstract was submitted to the Epidemics 2013 conference in Amsterdam, Netherlands. This model evaluated the impact of direct and indirect transmission terms. Duration of latent periods and contribution of early shedding status to long term outcome was evaluated as a part of the National Institute for Mathematical Biology within-host MAP modeling group. Contribution of stress events to shedding patterns was evaluated and a paper was in draft as of September 2013. Goal 3: The individual-based model of infection transmission of MAP was developed under the supervision of my mentor at Warwick University. This model was used to evaluate the impact of early shedding on endemicity in populations under intervention. This paper was in draft form as of September 2013.

PUBLICATIONS: 2012/09/01 TO 2013/08/31
1. Type: Journal Articles Status: Published Year Published: 2012 Citation: Salgado M, M Alfaro, F Salazar, E Troncoso, RM Mitchell, L Ramirez, A Naguil, P Zamorano, MT Collins (2013). Effect of soil slope on appearance of Mycobacterium avium subsp paratuberculosis in water running off grassland soil after contaminated slurry application. Applied Environmental Microbiology. Jun; 79(12): 3544-3552.
2. Type: Conference Papers and Presentations Status: Accepted Year Published: 2013 Citation: Casey L Cazer, Rebecca M Mitchell, Kellie M Cicconi-Hogan, Michael Gamroth, Roxanne M Richert, Pamela L Ruegg and Ynte H Schukken. Associations between Mycobacterium avium subsp. paratuberculosis antibodies in bulk tank milk, herd characteristics and management factors. American Association of Bovine Practitioners Annual Conference. September 19, 2013. Milwaukee, Wisconsin.
3. Type: Journal Articles Status: Submitted Year Published: 2013 Citation: Associations between Mycobacterium avium subsp. paratuberculosis antibodies in bulk tank milk, season of sampling and protocols for managing infected cows Casey L Cazer, Rebecca M Mitchell, Kellie M Cicconi-Hogan, Michael Gamroth, Roxanne M Richert, Pamela L Ruegg and Ynte H Schukken BMC Veterinary Research.
4. Type: Journal Articles Status: Submitted Year Published: 2013 Citation: nvasion and Transmission of Salmonella Kentucky in an Adult Dairy Herd Using Approximate Bayesian Computation. Zhao Lu, Rebecca M Mitchell, Rebecca L Smith, Jeffrey S Karns, Jo Ann S Van Kessel, David R Wolfgang, Ynte H Schukken and Yrjo T Grohn. BMC Veterinary Research.

PROGRESS: 2011/09/01 TO 2012/08/31
OUTPUTS: Initial results for mathematical modeling objectives (individual-based models), statistical analysis of adult animal shedding, and single-genome analysis were presented at the International Society for Veterinary Epidemiology and Economics conference August 2012, Maastricht, the Netherlands. During the course of research, two students worked with Dr. Mitchell: Lucia Favila-Humara, Borlaug Fellow Fall 2011 and Casey Cazer, Cornell Veterinary Student, Summer 2012. Dr. Favila-Humara will teach both the PCR for short sequence repeat typing and the single-genome analysis techniques to her colleagues in Mexico. Dr. Mitchell co-taught two short courses on mathematical modeling of endemic livestock diseases, one in Ithaca, NY and one at the International Society for Veterinary Epidemiology and Economics pre-conference seminar in Ghent, Belgium (August 2012). In addition, we held an informal weekly modeling seminar at Quality Milk Production Services, Fall-Winter 2011. PARTICIPANTS: Rebecca Mitchell is the PI/PD and only person who has received a stipend for working on this project as described in the initial project outline. Warwick University has provided facilities for Rebecca's time working with Graham Medley. Rebecca has trained a visiting scholar as well as a veterinary student over the course of this year. The visiting scholar focused on laboratory techniques (PCR and single genome analysis) and the veterinary student focused on statistical training. TARGET AUDIENCES: This far the target audience for knowledge gained through this project has been scientific colleagues at conferences as well as colleagues within the university. In addition, the preliminary results have been presented to the research and applied veterinary researchers at the University of Iceland and Keldur animal institute as well as to colleagues at the Harvard School of Public Health. Teaching of workshops concerning mathematical modeling has focused on both graduate students and veterinarians working in academia or government. PROJECT MODIFICATIONS: The only change to the project during this period was a shift from directed learning in the Cornell Center for Teaching Excellence to a focus on individual short course development and presentation. Cornell has shifted the focus of the Future Faculty Teaching Certificate graduate training program and post-doctoral researchers can no longer participate in the extra-curricular portion of the training. I completed the classroom teaching portion (ALS6015: Teaching in Higher Education) but have substituted for the remainder with multiple teaching experiences.

IMPACT: 2011/09/01 TO 2012/08/31
Through the first year of this project we had several knowledge changes. We were able to identify animals with multiple strains of MAP concurrently, recovered from the same tissue sample. These strains within the same sample were more closely related than strains from different samples, leading us to believe that there is within-host evolution of MAP strains. In addition, we were able to identify a relationship between exposure to shedding animals as adults and likelihood of becoming MAP positive at the level of tissue culture. This supports a hypothesis of adult susceptibility. In the individual-based models, we find multiple stable equilibria for infection which are supported by intervention efforts. This supports our original deterministic findings, but in herd sizes realistic for US dairy farms.

PUBLICATIONS: 2011/09/01 TO 2012/08/31
1. Accepted: Smith RL, YH Schukken, Z Lu, RM Mitchell, YT Grohn. (2012) Modeling infection dynamics of Mycobacterium bovis in US cattle herds and implications for control. Journal of the American Veterinary Medical Association.
2. Lu Z, YH Schukken, RL Smith, RM Mitchell, YT Grohn. (2012) Impact of imperfect Mycobacterium avium subspecies paratuberculosis vaccines in dairy herds: A mathematical modeling approach. Preventive Veterinary Medicine Aug 22. Epub ahead of print.
3. Heuer C, RM Mitchell, YH Schukken, Z Lu, C Verdugo, PR Wilson. (2012) Modelling transmission dynamics of paratuberculosis of red deer under pastoral farming conditions. Preventive Veterinary Medicine Sep 1; 106(1):64-74.