2014 Presentations

MinION™ device: latest advancements in next-generation DNA sequencing

Bethany A. Blank ¹, Scott J. Hebbring¹, Steven J. Schrodi¹, Zhan (Harold) Ye²
¹Center for Human Genetics, ²Biomedical Informatics Research Center
Research area: Genetics

Background: Advancements in next-generation sequencing (NGS) technologies are attempting to develop sequencing methods that are faster, more accurate, cheaper, and produce longer reads. The newest methods use single-molecule sequencing (SMS) to directly read DNA without the need of PCR amplification. As part of an early access program from Oxford Nanopore Technologies Ltd, we tested the capabilities and limitations of the MinION™ device, which uses nanopore technology for SMS, to determine the utility of the new technology.

Method: Seven DNA libraries were prepared from a lambda phage genome following the Oxford Nanopore Genomic DNA Sequencing Kit protocol. Libraries were loaded into the MinION Flow Cell and read using the MinKNOW™ software. Basecalling was performed using Metrichor™ software. Transformed data was then mapped to the reference sequence to determine quality, accuracy, and coverage.

Results: Quality of the MinION Flow Cells was quite varied ranging from greater than 100 to less than 400 active channels out of 512, greatly impacting the number of reads generated per library. Read length of fragments ranged from 5 to 109,320 bases, with an average of 4-5 kilobases compared to 200-300 bases using current NGS technologies. Basecalling quality averaged around a Phred score of 7 for bidirectional reads, which is significantly lower than current NGS technologies.

Conclusion: The MinION™ is a promising new DNA sequencing technology which after further development will likely be effective as a DNA sequencing device. The device’s ability to directly read DNA has many significant advantages, including eliminating the need for PCR and allowing more accurate de novo sequencing using longer DNA reads. However, issues relating to uncertain basecalling accuracy and quality, inconsistent flow cell quality, and development of compatible analysis tools must first be addressed.

Do Sleep disorders increase the risk of cardiovascular diseases? 

Colleen N. Bramwell, John R. Schmelzer, Richard L. Berg, Joseph J. Mazza, Jamie A. Boero, James K. Burmester. 
Clinical Research Center
Research area: Clinical Research

Background: Approximately 22 million Americans suffer from sleep apnea, characterized by pauses in breathing during sleep, resulting in reduction of blood oxygen levels. The most common treatment for sleep apnea is Continuous Positive Airway pressure (CPAP) that restores breathing and oxygen levels. Here we investigated whether there is an association between the severity of sleep apnea and cardiovascular disease.

Methods: We conducted a case-control study among adults who had undergone a sleep study from 2007 through 2013. We defined a case of cardiovascular disease as patients who experienced intervention (either Percutaneous Transluminal Coronary Angiogram or Coronary Artery Bypass Graft) or death subsequent to their sleep study. Two controls were matched to each case based on year of sleep study, age, gender and other clinical factors. Severity of sleep apnea was defined based on level of oxygen desaturation. Compliance with CPAP treatment was determined by two independent reviewers.

Results: Charts of 29 cases and 58 controls were reviewed. The mean age of patients was 61 years and 62% were male. Compliance with therapy was lower among cases than controls (69% vs 81%, odds ratio 0.47; 95% CI: 0.16-1.34). Minimum oxygen level was the severity measure showing the strongest association with events.

Conclusion: Our findings agree with previous research, suggesting that severity of sleep disturbance is a risk factor for cardiovascular events. Compliance with recommended treatment for sleep disorders may help reduce the risk of cardiovascular disease. Larger studies are needed to confirm these findings.

Patient Characteristics and Treatment of Fungal Infections in the Marshfield Clinic Health Care System

Matthew Buchholz, Jennifer Anderson, Jennifer Meece
Integrated Research and Development Laboratory
Research area: Clinical Research, Infectious Diseases

Background: Fungal diseases are an emerging threat to human health. As the population of immunocompromised individuals grows, an increased number of people are at risk of contracting a potentially fatal fungal infection. To better understand the patient characteristics associated with contracting fungal infections, we analyzed the demographics, comorbidities, and treatment of patients within the Marshfield Clinic Health Care System (MCHCS).

Methods: Cases were identified by the presence of a positive fungal culture with growth of Aspergillus spp., Blastomyces dermatitidis, Cryptococcus spp., or Histoplasma capsulatum and onset of symptoms from 2003-2013. A total of 906 patient charts were abstracted to identify demographics (gender, age, and race), presence or absence of 13 comorbidities, and antifungal treatment from the Combined Medical Record of the MCHCS. SAS version 9.3 was used for statistical analysis.

Results: Of the 906 cases, 604, 247, 38, and 17 represented Aspergillus spp., B. dermatitidis, Cryptococcus spp., and H. capsulatum infections, respectively. Combined cases ranged from 36-134 annually. All infection types showed propensity towards infecting males. B. dermatitidis infected individuals were younger than individuals with Aspergillus spp. or Cryptococcus spp. (p less than 0.001). Presence of at least one comorbidity ranged from 48-87% with hypertension, chronic obstructive pulmonary disease, diabetes, and cancer being the most common across all cases. The pulmonary tract was the most commonly infected site in all infection types. Pulmonary cases were significantly more likely to be hospitalized than non-pulmonary cases in all infections except H. capsulatum. The percent of patients receiving antifungal treatment varied from 32% of Aspergillus spp. cases to 98% of B. dermatitidis cases.

Conclusions: Our findings are consistent with other studies showing that fungal infections are most often pulmonary, accompanied by comorbidities, and more commonly infect males. Further studies are warranted to precisely determine risk factors of contracting a fungal infection.

Antibiotic Use in Patients with Medically-Attended Acute Respiratory Illness

James Bychinski, Maria Sundaram, Huong McLean, Edward Belongia
Center for Clinical Epidemiology and Population Health
Research area: Epidemiology 

Background: Acute respiratory illnesses (ARI) are a significant cause of morbidity. Outpatient ARIs are often treated with antibiotics regardless of etiology. We analyzed secular trends and identified factors associated with antibiotic prescribing among patients with medically attended viral ARI.

Methods: We used data from patients enrolled in influenza vaccine effectiveness studies from 2007-08 through 2012-13. ICD-9 codes were used to determine illness characteristics described as viral etiology. Secular trends in antibiotic prescribing were analyzed using Cochran-Armitage tests for trend. Associations between demographic/illness characteristics and antibiotic prescribing were assessed using chi-square tests. Logistic regression with stepwise selection was used to identify factors associated with antibiotic prescribing. Separate models were built for children and adults.

Results: Of 1189 children with viral illness, 239 (20%) received an antibiotic. Of 1852 adults with a viral illness, 939 (51%) received an antibiotic. For both adults and children, there was a statistically significant increasing trend in antibiotic use from 2008-2010; from 2011-2013, antibiotic use significantly decreased. In children, age, duration of illness, wheezing, study year, lack of confirmed influenza infection, and insurance type were associated with an increase in odds of antibiotic prescribing. In adults, age, duration of illness, study year and influenza vaccination were associated with increased odds of antibiotic prescribing. Clinical department in which the patient was seen, race, and gender were not associated with receipt of an antibiotic prescription in either children or adults.

Conclusion: Trends in antibiotic prescribing for viral diagnoses have not remained stable over time. Factors associated with antibiotic prescribing for viral illnesses differed between children and adults. More research is needed on the relationship between ICD-9 codes and illness etiology.

The Reliability of Clinical Adjudication Panels in the Longitudinal Study of Implantable Cardioverter Defibrillators

Charles E. Gaber, Robert Greenlee
Center for Clinical Epidemiology and Population Health
Research area: Epidemiology 

Background: Clinical trials have demonstrated the benefits of implantable cardioverter-defibrillators (ICDs) in the primary prevention of sudden cardiac death by delivering therapy for malignant ventricular arrhythmias. The Longitudinal Study of Implantable Cardioverter Defibrillators (LSICD) utilized clinical adjudication panels to determine the occurrence and appropriateness of ICD therapy outcomes in routine clinical practice settings. We characterized the reliability of these adjudication panels.

Methods: Study-eligible patients had to receive an ICD for the first time for the primary prevention of sudden cardiac death, have left ventricular systolic dysfunction, and be a patient in the seven participating health systems. All treated episodes for each patient were reviewed by a central physician panel, with a sample additionally reviewed by an external panel of experts. Kappa statistics and percent agreement were calculated to characterize the level of agreement between the local provider’s interpretation of therapy appropriateness, as documented in the medical record, and the final adjudicated determination of device therapy appropriateness. Similar calculations were made to assess inter-reviewer reliability in both panels and agreement between the final central panel and final external panel determinations.

Results: Kappa for agreement between the local provider determined appropriateness and the final adjudicated appropriateness was 0.79 (95% CI: 0.77, 0.81). Central panel inter-reviewer kappa statistics for the adjudicated variables ranged from 0.45 to 0.92. Specifically, kappa for central adjudicated appropriateness was 0.72 (95% CI: 0.69, 0.75). External panel inter-reviewer kappa statistics for the adjudicated variables ranged from 0.43 to 0.85. Kappa for external adjudicated appropriateness was 0.60 (95% CI: 0.54, 0.67). Kappa statistics for agreement between the final central adjudication and final external adjudication ranged from 0.19 to 0.95.

Conclusions: Agreement between and amongst panels is favorable. Adjudication panels are a reliable method of determining outcomes in the LSICD. Further studies can validate this finding in other study settings.

Developing a Unified Statistical Method for Analyzing Large-Scale Related and Unrelated Exome-Chip Data

Ethan Heinzen¹, Min He¹,²
¹Center for Human Genetics, and ²Biomedical Informatics Research Center
Research area: Genetics, Biomedical Informatics 

Background: When study populations in Genome-wide Association Studies (GWAS) contain both related and unrelated individuals, often the populations are split into two designs, family-based and case-control, which potentially reduces overall power. Data from exome-chip, an intermediate experiment between exome sequencing and genotyping arrays, was available on 10,016 individuals from the Personalized Medicine Research Project (PMRP) at the Marshfield Clinic Research Foundation. The population of the exome-chip data has a mixed structure, with over 40% of samples related. To analyze the full dataset appropriately, we developed an improved principal component analysis (iPCA) for related and unrelated exome-chip data that eliminates effects caused by population stratification (unrelated samples) and family structure (related samples).

Method: We excluded variants based on Hardy-Weinberg Equilibrium (<10^-6), minor allele frequency (<5%), high linkage disequilibrium (LD) regions, and high missingness (>10%). We identified related individuals using identity by descent (IBD), keeping as many unrelated individuals as possible. Next, we calculated eigenvectors of a covariance matrix based on unrelated samples. Finally, we calculated principal components for all samples by genetic difference from the mean multiplied by the calculated eigenvectors. We then tested associations with a chosen phenotype for each variant using a score-based test, incorporating covariates including BMI, smoking status, and the top 10 principal components to adjust for sample structure. Finally, we compared results with the Generalized Estimating Equation (GEE) model, which accounts for clusters in the same population.

Results: When implemented in R, the iPCA took 4.6 hours to calculate the top 10 principal components and 9.1 hours to calculate the top 100 principal components. The phenotype tested was asthma; the score-based test results are still pending.

Conclusion: The iPCA developed is promising, eliminating effects caused by population stratification and family structure, but is computationally intensive. Better and more efficient implementation of this iPCA is needed.

Dental Hygienists’ Information Needs Assessment 

David Kirkhoff, Kelsey Schwei, Amit Acharya
Institute for Oral and Systemic Health, Biomedical Informatics Research Center
Research area: Dental Informatics 

Background: Similar to dentists, dental hygienists (DHs) are involved in retrieving and documenting patient data as part of their decision-making process. Switching between medical, dental, and medication histories, radiographs, and hard tissue charting complicates this process. The archaic periodontal chart must be updated to include information crucial to the DHs’ needs. The objective of this study was to assess the workflow and information needs of DHs and to aid in the design of a smart periodontal charting interface.
Method: Two methodologies were used: cognitive task analysis (CTA) and ethnographic observations. For the CTA sessions, three fictitious patient profiles were created and reviewed by a dentist. The DHs were presented with pieces of information regarding the fictitious patient and were asked to think aloud about what pieces of information they would like to review in the decision-making process. The sessions were audio and video recorded, transcribed, and coded as the following activities: information retrieval, decision making, information processing, and other. The notes taken during the ethnographic observation sessions were then analyzed for themes.

Results: Four DHs participated in both the CTA and observation sessions. For CTA sessions, medical history was reviewed first 8/12 times. The participants devoted an average of 41.3% (SD =7.2%) to information retrieval, 36.2% (SD =7.2%) to decision making, 9.2% (SD =4.3%) to information processing, and 12.9% (SD =13.2%) to other. During the observation sessions, all participants used the universal numbering system to record probe depths using the keyboard. Participants often navigated away from the periodontal chart to consult radiographs during the patient exam.

Conclusion: DHs devoted a significant amount of effort to retrieving and reviewing patient information and spent considerably less time processing that information. Including a link to radiographs and medical histories in the periodontal chart would increase efficiencies in workflow.

Development of Oral Cancer Risk Assessment Tool: An Exploratory Study

Krista J. Koehler, Neel A. Shimpi, Amit Acharya, Harshad Hedge, Gary Pack
Institute for Oral and Systemic Health, Biomedical Informatics Research Center
Research area: Dental Informatics 

Background: Currently, there is no standard procedure for identifying patient risk of oral cancer. A combination of several etiological factors such as smoking, alcohol, oral Human Papilloma Virus and others have been associated with the risk of oral cancer. With advanced studies showing multiple etiological factors with synergistic effects acting on oral cancer development, it becomes difficult for health care providers to determine patient risk of oral cancer. We explored the design of an oral cancer risk assessment tool (OCRAT) based on MATLAB Artificial Neural Network (ANN).

Methods: Retrospective data was collected from the Marshfield Clinic data warehouse from 01/01/1979 to 06/06/2014. Structured data was filtered by applying inclusion and exclusion criteria to develop data sets of 300 cases and 300 controls, which were trained, tested, and validated using MATLAB. Only White, Non-Hispanic patients were included. The ANN had a total of 7 input neurons that included patient age, gender, white lesions, red lesions, submucous fibrosis, current smoker former smoker, 4 hidden nodes, and 2 outputs of either high or low risk of developing oral cancer. Performance function of the OCRAT prototype was analyzed on the validation set using a confusion matrix and mean square error (MSE).

Results: The validation set demonstrated a high specificity of 95.6%, sensitivity of 100%, while precision and recall were 95.2% and 100% respectively. The best epoch was at 11 with MSE of 0.09. Overall accuracy of MATLAB performance was 97.7%.

Conclusion: Progress designing a prototypical tool that had reasonable accuracy was made and provided an opportunity to understand the use of MATLAB and ANN methodology for the purpose of risk prediction. Further data mining through clinical notes may provide more information for input nodes including other known risk factors for oral cancer including alcohol consumption, thereby increasing the efficacy of OCRAT prototype.

Agriculture and a Culture of Safety: Mapping Spheres of Influence

Jessica ME Perkins MPH, Casper G Bendixsen PhD
National Farm Medicine Center
Research area: Agriculture Health and Safety, Anthropology 

Background: Agriculture is one of the most dangerous occupational sectors, with high rates of injury/illness and death. Agricultural health and safety professionals have had limited success in delivering their programs and messages to farmers. However, other individuals have earned farmers’ trust and influence their decision-making. This pilot project will design and use a pile sort instrument, an ethnographic research method that gathers both qualitative and quantitative data, to investigate and describe the relationship between agricultural bankers and dairy farmers in central Wisconsin.

Methods: Agricultural bankers and dairy farmers participated in one-hour interviews where questions were asked using a pile sort that consisted of 32 cards. Each card had an individual/institution written on it, with whom farmers and bankers interact. The same set of cards was used for both groups with the questions being complementary. Demographic questions were asked to characterize their dairy/portfolio. Lead-in questions were used to assess participants’ perspectives on farm safety as part of a successful business plan. Participants sorted cards based on a modified 5-point Likert scale to assess the relationship with the farmer, their knowledge of farm safety, and the amount of trust afforded to them by the farmer.

Results: Preliminary ethnographic evidence shows dairy farmers and agricultural bankers tend to agree that spouses, immediate family, and herdsmen are on-farm trusted advisors in farm safety. Off-farm examples were also agreed upon: agricultural health and safety professionals and veterinarians. Moreover, both agree that these individuals are trusted enough to make immediate changes based on their recommendations.

Conclusions: Both agricultural bankers and dairy farmers believe farm safety is important in the financial success of a dairy farm. There is potential for agricultural health and safety experts to utilize these individuals to help design and implement farm safety programs, build rapport, and incentivize safety through other important institutions.

Association between Genetic Variants in ARMS2 and the Risk of Age-related Macular Degeneration

Ran Zhao¹, Scott J Hebbring¹, Zhan (Harold) Ye²
¹Center for Human Genetics, ²Biomedical Informatics Research Center
Research area: Genetics 

Background: Age-related Macular Degeneration (AMD) is the leading cause of blindness in adults. It is well known that AMD is associated with multiple genetic markers in multiple genes on chromosome 10, including a common nonsynonymous single nucleotide polymorphism (SNP) (rs10490924) in ARMS2. However, the causative SNP for AMD in this region is difficult to identify because of extensive linkage disequilibrium. A nonsense SNP (rs2736911) in ARMS2 has been identified and may be hypothesized to reduce ARMS2 function. To determine if variants in ARMS2 may predict AMD, we genotyped these two coding variants in AMD cases and controls.

Method: We genotyped 865 cases and 1496 controls in a subset of patients within the Personalized Medicine Research Project using an allele specific PCR. Two fluorescently labeled and two unlabeled primers of different lengths were used to amplify alleles corresponding to either the wild-type or the variant allele for both coding SNPs in ARMS2. The frequency of each haplotype and diplotype were compared; p-values were obtained from a chi-square test.

Results: The frequency of SNP rs10490924 was enriched in cases compared to controls as expected (OR=1.5; 95% CI: 1.3, 1.7; p <0.001); individuals who were compound heterozygous for both coding SNPs had a similar OR compared to individuals who were homozygous for SNP rs10490924, suggesting SNP rs2736911 may modify the apparent association of SNP rs10490924.

Conclusion: This study is consistent with previous studies in that SNP rs10490924 was associated with AMD. Furthermore, SNP rs2736911 may modify the effect of SNP rs10490924, which suggests that SNP rs10490924 may be the causative SNP for AMD. However, further studies are still needed in order to identify the functional relationship between ARMS2 and AMD.