The Mayo Clinic Biobank: a building block for individualized medicine

Differential Serum Levels of CACNA1C, Circadian Rhythm and Stress Response Molecules in Subjects with Bipolar Disorder: Associations with Genetic and Clinical Factors

Background

Many patients with bipolar disorder (BD) do not respond to or have difficulties tolerating lithium and/or other mood stabilizing agents. There is a need for personalized treatments based on biomarkers in guiding treatment options. The calcium voltage-gated channel CACNA1C is a promising candidate for developing personalized treatments. CACNA1C is implicated in BD by genome-wide association studies and several lines of evidence suggest that targeting L-type calcium channels could be an effective treatment strategy. However, before such individualized treatments can be pursued, biomarkers predicting treatment response need to be developed.

Methods

As a first step in testing the hypothesis that CACNA1C genotype is associated with serum levels of CACNA1C, we conducted ELISA measures on serum samples from 100 subjects with BD and 100 control subjects.

Results

We observed significantly higher CACNA1C (p<0.01) protein levels in subjects with BD. The risk SNP (rs11062170) showed functional significance as subjects homozygous for the risk allele (CC) had significantly greater CACNA1C protein levels compared to subjects with one (p=0.013) or no copies (p=0.009). We observed higher somatostatin (SST) (p<0.003) protein levels and lower levels of the clock protein ARTNL (p<0.03) and stress signaling factor corticotrophin releasing hormone (CRH) (p<0.001) in BD. SST and PER2 protein levels were associated with both alcohol dependence and lithium response.

Conclusions

Our findings represent the first evidence for increased serum levels of CACNA1C in BD. Along with altered levels of SST, ARNTL, and CRH our findings suggest CACNA1C is associated with circadian rhythm and stress response disturbances in BD.

Clinical Trials for Special Populations: Children, Older Adults, and Rare Diseases

Research cannot maximize population health unless it improves health for all members of the public, including special populations such as children, older adults, and people living with rare diseases. Each of these categories require special considerations when planning and performing clinical trials, and common threads of ethical conduct of research in vulnerable populations appear throughout. In this review, definitions of each of the three categories of special population (children, older adults, and rare diseases) are discussed in terms of US research regulations, the unique challenges to conducting clinical trials for these special populations, critical ethical issues, and opportunities for innovative ways to design and operationalize clinical trials in special populations. Additional critical attention is focused on factors that influence the generalizability of study results to reduce health disparities, as well as the importance of community engagement and advocacy groups that can help to educate potential trial participants of the benefits of clinical trial participation.

Strategies for Implementing Machine Learning Algorithms in the Clinical Practice of Radiology

Despite recent advancements in machine learning (ML) applications in health care, there have been few benefits and improvements to clinical medicine in the hospital setting. To facilitate clinical adaptation of methods in ML, this review proposes a standardized framework for the step-by-step implementation of artificial intelligence into the clinical practice of radiology that focuses on three key components: problem identification, stakeholder alignment, and pipeline integration. A review of the recent literature and empirical evidence in radiologic imaging applications justifies this approach and offers a discussion on structuring implementation efforts to help other hospital practices leverage ML to improve patient care. Clinical trial registration no. 04242667 © RSNA, 2024 Supplemental material is available for this article.

High-Dimensional Clustering of 4000 Irritable Bowel Syndrome Patients Reveals Seven Distinct Disease Subsets

BACKGROUND AND AIMS

Irritable bowel syndrome (IBS) is a pain disorder classified by bowel habits, disregarding other factors that may influence the clinical course. The aim of this study was to determine if IBS patients can be clustered based on clinical, dietary, lifestyle, and psychosocial factors.

METHODS

Between 2013 and 2020, the Mayo Clinic Biobank surveyed and received 40,291 responses to a questionnaire incorporating Rome III criteria. Factors associated with IBS were determined and latent class analysis, a model-based clustering, was performed on IBS cases.

RESULTS

We identified 4021 IBS patients (mean 64 years; 75% women) and 12,063 controls. Using 26 variables separating cases from controls, the optimal clustering revealed 7 latent clusters. These were characterized by perceived health impairment (moderate or severe), psychoneurological factors, and bowel dysfunction (diarrhea or constipation predominance). Health impairment clusters demonstrated more pain, with the severe cluster also having more psychiatric comorbidities. The next 3 clusters had unique enrichment of psychiatric, neurological, or both comorbidities. The bowel dysfunction clusters demonstrated less abdominal pain, with diarrhea cluster most likely to report pain improvement with defecation. The constipation cluster had the highest exercise score and consumption of fruits, vegetables, and alcohol. The distribution of clusters remained similar when Rome IV criteria were applied. Physiologic tests were available on a limited subset (6%), and there were no significant differences between clusters.

CONCLUSIONS

In this cohort of older IBS patients, 7 distinct clusters were identified demonstrating varying degrees of gastrointestinal symptoms, comorbidities, dietary, and lifestyle factors. Further research is required to assess whether these unique clusters could be used to direct clinical trials and individualize patient management.

Type 2 Diabetes Remission in Patients with Heterozygous Variants in the Leptin-Melanocortin Pathway after Roux-en-Y Gastric Bypass: A Matched Case-Control Study

BACKGROUND

Roux-en-Y gastric bypass (RYGB) is associated with a high rate of type 2 diabetes (T2D) remission. Carriers of heterozygous variants in the leptin-melanocortin pathway (LMP) are more likely to experience weight recurrence after RYGB. Our aim was to investigate if carrier status and associated weight regain affects the rate of T2D remission after RYGB.

METHODS

Carriers of LMP variants with a diagnosis of T2D prior to RYGB (N = 16) were matched to non-carriers (N = 32) based on sex, age, and BMI. We assessed for post-operative T2D remission status post-surgery on a yearly basis, for up to 15 years. Our primary endpoint was the proportion of patients achieving T2D remission at 1 year. We conducted a survival analysis for all patients that achieved remission at least at one time-point to evaluate for maintenance of T2D remission by using a log-rank test.

RESULTS

Both carriers and non-carriers had similar baseline and procedural characteristics. The proopiomelanocortin gene in the LMP pathway had the most variants (n = 5, 31%). Carriers had a lower total body weight loss percentage at nadir (28.7% ± 6.9) than non-carriers (33.7% ± 8.8, p = 0.04). The proportion of patients achieving T2D remission at 1 year was 68.8% for carriers and 71.9% for non-carriers (p = 1.0). Survival curves for maintenance of first remission were similar for both groups (p = 0.73), with a median survival of 8 years for both carriers and non-carriers.

CONCLUSIONS

Despite inferior weight loss outcomes at nadir, carriers had similar T2D remission rates when compared to non-carriers. Weight-independent metabolic benefits of RYGB might contribute to this observation.

Cohort study examining associations between ceramide levels and risk of multimorbidity among persons participating in the Mayo Clinic Biobank

OBJECTIVE

Ceramides have been associated with several ageing-related conditions but have not been studied as a general biomarker of multimorbidity (MM). Therefore, we determined whether ceramide levels are associated with the rapid development of MM.

DESIGN

Retrospective cohort study.

SETTING

Mayo Clinic Biobank.

PARTICIPANTS

1809 persons in the Mayo Clinic Biobank ≥65 years without MM at the time of enrolment, and with ceramide levels assayed from stored plasma.

PRIMARY OUTCOME MEASURE

Persons were followed for a median of 5.7 years through their medical records to identify new diagnoses of 20 chronic conditions. The number of new conditions was divided by the person-years of follow-up to calculate the rate of accumulation of new chronic conditions.

RESULTS

Higher levels of C18:0 and C20:0 were associated with a more rapid rate of accumulation of chronic conditions (C18:0 z score RR: 1.30, 95% CI: 1.10 to 1.53; C20:0 z score RR: 1.26, 95% CI: 1.07 to 1.49). Higher C18:0 and C20:0 levels were also associated with an increased risk of hypertension and coronary artery disease.

CONCLUSIONS

C18:0 and C20:0 were associated with an increased risk of cardiometabolic conditions. When combined with biomarkers specific to other diseases of ageing, these ceramides may be a useful component of a biomarker panel for predicting accelerated ageing.

The importance of social activity to risk of major depression in older adults

BACKGROUND

Several social determinants of health (SDoH) have been associated with the onset of major depressive disorder (MDD). However, prior studies largely focused on individual SDoH and thus less is known about the relative importance (RI) of SDoH variables, especially in older adults. Given that risk factors for MDD may differ across the lifespan, we aimed to identify the SDoH that was most strongly related to newly diagnosed MDD in a cohort of older adults.

METHODS

We used self-reported health-related survey data from 41 174 older adults (50-89 years, median age = 67 years) who participated in the Mayo Clinic Biobank, and linked ICD codes for MDD in the participants' electronic health records. Participants with a history of clinically documented or self-reported MDD prior to survey completion were excluded from analysis (N = 10 938, 27%). We used Cox proportional hazards models with a gradient boosting machine approach to quantify the RI of 30 pre-selected SDoH variables on the risk of future MDD diagnosis.

RESULTS

Following biobank enrollment, 2073 older participants were diagnosed with MDD during the follow-up period (median duration = 6.7 years). The most influential SDoH was perceived level of social activity (RI = 0.17). Lower level of social activity was associated with a higher risk of MDD [hazard ratio = 2.27 (95% CI 2.00-2.50) for highest v. lowest level].

CONCLUSION

Across a range of SDoH variables, perceived level of social activity is most strongly related to MDD in older adults. Monitoring changes in the level of social activity may help identify older adults at an increased risk of MDD.

Comparison of Demographic and Clinical Features of Bipolar Disorder in Persons of African and European Ancestry

AIM

This study quantified and compared demographic and clinical features of bipolar disorder (BD) in persons of African ancestry (AA) and European ancestry (EUR).

METHODS

Participants enrolled in the Mayo Clinic Bipolar Biobank from 2009 to 2015. The structured clinical interview for DSM-IV was used to confirm the diagnosis of BD, and a questionnaire was developed to collect data on the clinical course of illness. Descriptive statistics and bivariate analyses were completed to compare AA versus EUR participants. Subsequently, clinical outcomes were compared between AA and EUR participants using linear regression for continuous outcomes or logistic regression for binary outcomes while controlling for differences in age, sex, and recruitment site.

RESULTS

Of 1865 participants enrolled in the bipolar biobank, 65 (3.5%) self-identified as AA. The clinical phenotype for AA participants, in comparison to EUR participants, was more likely to include a history of PTSD (39.7% vs. 26.2%), cocaine use disorder (24.2% vs. 11.9%), and tardive dyskinesia (7.1% vs. 3%).

CONCLUSION

The low rate of AA enrollment is consistent with other genetic studies. While clinical features of bipolar disorder are largely similar, this study identified differences in rates of trauma, substance use, and tardive dyskinesia that may represent health disparities in bipolar patients of African ancestry. Future bipolar biomarker studies with larger sample sizes focused on underrepresented populations will provide greater ancestry diversity in genomic medicine with greater applicability to diverse patient populations, serving to inform health care policies to address disparities in bipolar disorder.

Effects of Heterozygous Variants in the Leptin-Melanocortin Pathway on Transoral Outlet Reduction After Roux-en-Y Gastric Bypass: A Case-Control Study and Review of Literature

BACKGROUND

Transoral outlet reduction (TORe) is a safe and effective technique for management of weight regain (WR) after Roux-en-Y Gastric Bypass (RYGB). Carriers of a heterozygous variant in the leptin melanocortin pathway (LMP) have been shown to be at high risk for WR in the mid- and long-term after RYGB. Our case series includes four patients with heterozygous LMP variants and presents novel data on their weight loss after TORe.

METHODS

We performed a retrospective study of the Mayo Clinic Biobank and identified adult participants who had been genotyped and found to have or do not have a heterozygous variant in the LMP ("carriers" vs "non-carriers", respectively) and had undergone a TORe procedure. TBWL% at 1, 3, 6, 9, and 12 months ± 15 days were calculated based on baseline weight at TORe procedure.

RESULTS

A total of 14 patients were included in the analysis: four patients (mean age 51.0 [5.2] years, 100% females, body mass index [BMI] 40.5 [8.7] kg/m²) with LMP variant and 10 non-carriers (age 55.4 [15.3] years, 90% females, BMI 37.3 [7.7] kg/m²). There were no baseline differences between carriers and non-carriers at time of TORe procedure. After TORE, carriers lost less weight when compared to non-carriers at 3, 6, 9, and 12 months. The difference at 12 months was statistically significant (1.6 vs 12.3%; p = 0.03).

CONCLUSIONS

Patients with a LMP variant and that underwent RYGB showed decreased weight loss after undergoing TORe. Further and larger studies are needed to comprehend the effect of TORe on patients with LMP variants.

Association between work physical activity, dietary factors, and risk of rheumatoid arthritis

OBJECTIVE

We aimed to determine the association of physical activity and dietary factors on RA risk.

METHODS

This case-control study within the Mayo Clinic Biobank matched incident RA cases (two codes plus disease-modifying anti-rheumatic drug, PPV 95%) to controls 1:3 on age, sex, and recruitment year/location. A baseline questionnaire assessed activity and dietary exposures. Logistic regression models calculated adjusted odds ratios (aOR) with 95% confidence intervals (CI) of RA for each of 45 activity/dietary exposures.

RESULTS

We identified 212 incident RA cases and 636 controls (mean age 64, 70% female). Active work physical activity was associated with elevated risk of RA (aOR 3.00, 95% CI 1.58-5.69 vs. sedentary); leisure activity was not (aOR 0.96, 95% CI 0.64-1.42 sedentary vs. active). Three or more servings high-fat food and 5+ servings fruits/vegetables daily showed non-significant associations with RA (aOR 1.22, 95% CI 0.74-2.00 vs. 0-1 time; aOR 0.75, 95% CI 0.51-1.11 vs. 0-3 times), especially in sensitivity analyses with at least five years between questionnaire and RA (aOR 1.80, 95% CI 0.69-4.71; aOR 0.54, 95% CI 0.27-1.08). Alcohol binging was not associated with RA risk (aOR 1.28, 95% CI 0.56-2.96). Finally, sensitivity (versus primary) analyses showed a nonsignificant increase in RA risk for most vitamins and supplements.

CONCLUSION

Active work physical activity and some nutritional profiles (increased high-fat, reduced fruit/vegetable consumption) may be associated with increased risk of RA. Confirmatory studies are needed.

Employment Characteristics and Risk of Hospitalization Among Older Adults Participating in the Mayo Clinic Biobank

Objective

To determine the relationship between characteristics of employment and future hospitalization in older adults.

Patients and Methods

We conducted a survey of adults aged 65 years or older participating in the Mayo Clinic Biobank. Using a frequency-matched, case-control design, we compared patients who were hospitalized within 5 years of biobank enrollment (cases) with those who were not hospitalized (controls). We assessed the duration of work, age at first job, number of jobs, disability, retirement, and reasons for leaving work. We performed logistic regression analysis to assess the association of these factors with hospitalization, accounting for age, sex, comorbid conditions, and education level.

Results

Among 3536 participants (1600 cases and 1936 controls; median age, 68.5 years; interquartile range, 63.4-73.9 years), cases were older, more likely to be male, and had lower education levels. Comorbid illnesses had the largest association with hospitalization (odds ratio [OR], 4.09; 95% CI, 3.37-4.97 [highest vs lowest quartile]). On adjusted analyses, odds of hospitalization increased with the presence of disability (OR, 1.31; 95% CI, 1.01-1.69) and decreased with having 1 or 2 lifetime jobs vs no employment (OR, 0.77; 95% CI, 0.60-1.00). The length of work, furlough, age of retirement, childcare issues, and reasons for leaving a job were not associated with hospitalization.

Conclusion

This study reports an association between disability during work and hospitalization. On the basis of our findings, it may be important to obtain a more detailed work history from patients because it may provide further insight into their future health.

Progression and survival of MBL: a screening study of 10 139 individuals

Monoclonal B-cell lymphocytosis (MBL) is a common hematological premalignant condition that is understudied in screening cohorts. MBL can be classified into low-count (LC) and high-count (HC) types based on the size of the B-cell clone. Using the Mayo Clinic Biobank, we screened for MBL and evaluated its association with future hematologic malignancy and overall survival (OS). We had a two-stage study design including discovery and validation cohorts. We screened for MBL using an eight-color flow-cytometry assay. Medical records were abstracted for hematological cancers and death. We used Cox regression to evaluate associations and estimate hazard ratios and 95% confidence intervals (CIs), adjusting for age and sex. We identified 1712 (17%) individuals with MBL (95% LC-MBL), and the median follow-up time for OS was 34.4 months with 621 individuals who died. We did not observe an association with OS among individuals with LC-MBL (P = .78) but did among HC-MBL (hazard ratio, 1.8; 95% CI, 1.1-3.1; P = .03). Among the discovery cohort with a median of 10.0 years follow-up, 31 individuals developed hematological cancers with two-thirds being lymphoid malignancies. MBL was associated with 3.6-fold risk of hematological cancer compared to controls (95% CI, 1.7-7.7; P < .001) and 7.7-fold increased risk for lymphoid malignancies (95% CI:3.1-19.2; P < .001). LC-MBL was associated with 4.3-fold risk of lymphoid malignancies (95% CI, 1.4-12.7; P = .009); HC-MBL had a 74-fold increased risk (95% CI, 22-246; P < .001). In this large screening cohort, we observed similar survival among individuals with and without LC-MBL, yet individuals with LC-MBL have a fourfold increased risk of lymphoid malignancies. Accumulating evidence indicates that there are clinical consequences to LC-MBL, a condition that affects 8 to 10 million adults in the United States.

Design and validation of a FHIR-based EHR-driven phenotyping toolbox

OBJECTIVES

To develop and validate a standards-based phenotyping tool to author electronic health record (EHR)-based phenotype definitions and demonstrate execution of the definitions against heterogeneous clinical research data platforms.

MATERIALS AND METHODS

We developed an open-source, standards-compliant phenotyping tool known as the PhEMA Workbench that enables a phenotype representation using the Fast Healthcare Interoperability Resources (FHIR) and Clinical Quality Language (CQL) standards. We then demonstrated how this tool can be used to conduct EHR-based phenotyping, including phenotype authoring, execution, and validation. We validated the performance of the tool by executing a thrombotic event phenotype definition at 3 sites, Mayo Clinic (MC), Northwestern Medicine (NM), and Weill Cornell Medicine (WCM), and used manual review to determine precision and recall.

RESULTS

An initial version of the PhEMA Workbench has been released, which supports phenotype authoring, execution, and publishing to a shared phenotype definition repository. The resulting thrombotic event phenotype definition consisted of 11 CQL statements, and 24 value sets containing a total of 834 codes. Technical validation showed satisfactory performance (both NM and MC had 100% precision and recall and WCM had a precision of 95% and a recall of 84%).

CONCLUSIONS

We demonstrate that the PhEMA Workbench can facilitate EHR-driven phenotype definition, execution, and phenotype sharing in heterogeneous clinical research data environments. A phenotype definition that integrates with existing standards-compliant systems, and the use of a formal representation facilitates automation and can decrease potential for human error.

Effects of Heterozygous Variants in the Leptin-Melanocortin Pathway on Roux-en-Y Gastric Bypass Outcomes: a 15-Year Case-Control Study

INTRODUCTION

Heterozygous variants in the leptin-melanocortin pathway are associated with obesity. However, their effect on the long-term outcomes after Roux-en-Y gastric bypass (RYGB) is still unknown.

METHODS

In this matched case-control study, 701 participants from the Mayo Clinic Biobank with a history of RYGB were genotyped. Sixty-three patients had a heterozygous variant in the leptin-melanocortin pathway. After excluding patients with potential confounders, carriers were randomly matched (on sex, age, body mass index [BMI], and years since surgery) with two non-carrier controls. The electronic medical record of carriers and matched non-carriers was reviewed for up to 15 years after RYGB.

RESULTS

A total of 50 carriers and 100 matched non-carriers with a history of RYGB were included in the study. Seven different genes (LEPR, PCSK1, POMC, SH2B1, SRC1, MC4R, and SIM1) in the leptin-melanocortin pathway were identified. At the time of surgery, the mean age was 50.8 ± 10.6 years, BMI 45.6 ± 7.3 kg/m², and 79% women. There were no differences in postoperative years of follow-up, Roux limb length, or gastric pouch size between groups. Fifteen years after RYGB, the percentage of total body weight loss (%TBWL) in carriers was - 16.6 ± 10.7 compared with - 28.7 ± 12.9 in non-carriers (diff = 12.1%; 95% CI, 4.8 to 19.3) and the percentage of weight regain after maximum weight loss was 52.7 ± 29.7 in carriers compared with 29.8 ± 20.7 in non-carriers (diff = 22.9%; 95% CI, 5.3 to 40.5). The nadir %TBWL was lower - 32.1 ± 8.1 in carriers compared with - 36.8 ± 10.4 in non-carriers (diff = 4.8%; 95% CI 1.8 to 7.8).

CONCLUSIONS

Carriers of a heterozygous variant in the leptin-melanocortin pathway have a progressive and significant weight regain in the mid- and long-term after RYGB. Genotyping patients experiencing significant weight regain after RYGB could help implement multidisciplinary and individualized weight loss interventions to improve weight maintenance after surgery.

Genetic Determinants in Airways Obstructive Diseases: The Case of Asthma Chronic Obstructive Pulmonary Disease Overlap

Genome-wide association studies (GWAS) of asthma and chronic obstructive pulmonary disease (COPD) with ever-increasing sample sizes have found multiple genetic loci associated with either disease. However, there are few intersecting loci between asthma and COPD. GWAS specifically focused on asthma-COPD overlap (ACO) have been limited by smaller sample sizes and the lack of a consistent definition of ACO that has also hampered clinical and epidemiologic studies. Other genomic techniques, such as gene expression profiling, are feasible with smaller sample sizes. Genetic analyses of objective measures of airway reactivity and allergy/T2 inflammation biomarkers in COPD studies may be another strategy to overcome limitations in ACO definitions.

What can data trusts for health research learn from participatory governance in biobanks?

New models of data governance for health data are a focus of growing interest in an era of challenge to the social licence. In this article, we reflect on what the data trust model, which is founded on principles of participatory governance, can learn from experiences of involving and engagement of members of the public and participants in the governance of large-scale biobanks. We distinguish between upstream and ongoing governance models, showing how they require careful design and operation if they are to deliver on aspirations for deliberation and participation. Drawing on this learning, we identify a set of considerations important to future design for data trusts as they seek to ensure just, proportionate and fair governance. These considerations relate to the timing of involvement of participants, patterns of inclusion and exclusion, and responsiveness to stakeholder involvement and engagement. We emphasise that the evolution of governance models for data should be matched by a commitment to evaluation.

Fruit, Vegetable, and Physical Activity Guideline Adherence and Metabolic Syndrome in El Banco por Salud

Adherence to dietary and physical activity recommendations has been associated with reductions in morbidity and mortality. The association between baseline adherence to fruit, vegetable, and physical activity guidelines and metabolic syndrome (MetS) in El Banco por Salud (El Banco) was examined. El Banco is a wellness biobank for Latino individuals affiliated with partnered Federally Qualified Health Centers in southern Arizona. Study participants (n = 972) were 65% female, 62.3% foreign-born, 56.3% obese, 29.2% food insecure, and with an average age of 51.3 years. Adherence scores were developed using baseline questionnaires for fruits and vegetable consumption and self-reported physical activity. Adherence was low in those fully meeting guidelines for fruit, vegetable, and physical activity at 14.6%, 37.5%, and 23.5%, respectively. Roughly 65% (n = 630) had ≥3 cardiometabolic risk factors. Large waist circumference was the most prevalent risk factor at 77.9%. Adherence to physical activity recommendations differed by MetS status with 32.8% without MetS reporting ≥150 min of physical activity per week compared to 18.5% in those with MetS (p < 0.001). There were no significant associations with adherence to any guidelines and MetS in the fully adjusted model. Overall, in this sample guideline adherence was low and the cardiometabolic risk factors prevalence was high.

Whole Genome Sequencing in Hypoplastic Left Heart Syndrome

Hypoplastic left heart syndrome (HLHS) is a genetically complex disorder. Whole genome sequencing enables comprehensive scrutiny of single nucleotide variants and small insertions/deletions, within both coding and regulatory regions of the genome, revolutionizing susceptibility-gene discovery research. Because millions of rare variants comprise an individual genome, identification of alleles linked to HLHS necessitates filtering algorithms based on various parameters, such as inheritance, enrichment, omics data, known genotype–phenotype associations, and predictive or experimental modeling. In this brief review, we highlight family and cohort-based strategies used to analyze whole genome sequencing datasets and identify HLHS candidate genes. Key findings include compound and digenic heterozygosity among several prioritized genes and genetic associations between HLHS and bicuspid aortic valve or cardiomyopathy. Together with findings of independent genomic investigations, MYH6 has emerged as a compelling disease gene for HLHS and other left-sided congenital heart diseases.

Prevalence of Heavy Chain MGUS by Race and Family History Risk Groups Using a High Sensitivity Screening Method

The mass spectrometry assay found over threefold numbers of individuals with MGUS than gel-based assays across 3 risk groups.

Relative differences in MGUS using the sensitive mass spectrometry assay were similar by race, family history, and age as prior MGUS studies.

Mass-spectrometry (MS) assays detect lower levels of monoclonal proteins and result in earlier detection of monoclonal gammopathy of undetermined significance (MGUS). We examined heavy chain MGUS prevalence using matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) MS among 3 risk groups, ages 50 or older: 327 African Americans (AA) and 1223 European Americans (EA) from a clinical biobank and 1093 unaffected first-degree relatives (FDR) of patients with hematologic disorders. Age- and sex-adjusted prevalence rates were directly standardized to 2010 United States population. Prevalence ratios were estimated for comparisons of AA and FDR to the EA group using the Poisson distribution. Results were also compared with population-based prevalence using conventional gel-based methods. Risk groups had similar sex and age distributions. MALDI-TOF MGUS prevalence was higher in the AA (16.5% [95% confidence interval (CI), 12.2%, 20.8%]) and FDR (18.3% [95% CI, 16.6%, 21.6%]) than in EA (10.8% [95% CI, 8.8%, 12.7%]), translating to prevalence ratios of 1.73 (95% CI, 1.31, 2.29) and 1.90 (95% CI, 1.55, 2.34), respectively. MALDI-TOF EA prevalence was over threefold higher than conventional estimates but showed similar age trends. Thus, the MALDI-TOF assay found greater numbers with MGUS but similar relative differences by race, family history, and age as prior studies.

A hybrid model to identify fall occurrence from electronic health records

INTRODUCTION

Falls are a leading cause of unintentional injury in the elderly. Electronic health records (EHRs) offer the unique opportunity to develop models that can identify fall events. However, identifying fall events in clinical notes requires advanced natural language processing (NLP) to simultaneously address multiple issues because the word "fall" is a typical homonym.

METHODS

We implemented a context-aware language model, Bidirectional Encoder Representations from Transformers (BERT) to identify falls from the EHR text and further fused the BERT model into a hybrid architecture coupled with post-hoc heuristic rules to enhance the performance. The models were evaluated on real world EHR data and were compared to conventional rule-based and deep learning models (CNN and Bi-LSTM). To better understand the ability of each approach to identify falls, we further categorize fall-related concepts (i.e., risk of fall, prevention of fall, homonym) and performed a detailed error analysis.

RESULTS

The hybrid model achieved the highest f1-score on sentence (0.971), document (0.985), and patient (0.954) level. At the sentence level (basic data unit in the model), the hybrid model had 0.954, 1.000, 0.988, and 0.999 in sensitivity, specificity, positive predictive value, and negative predictive value, respectively. The error analysis showed that that machine learning-based approaches demonstrated higher performance than a rule-based approach in challenging cases that required contextual understanding. The context-aware language model (BERT) slightly outperformed the word embedding approach trained on Bi-LSTM. No single model yielded the best performance for all fall-related semantic categories.

CONCLUSION

A context-aware language model (BERT) was able to identify challenging fall events that requires context understanding in EHR free text. The hybrid model combined with post-hoc rules allowed a custom fix on the BERT outcomes and further improved the performance of fall detection.

Sauver J, Sohn S. Ascertainment of Delirium Status Using Natural Language Processing From Electronic Health Records

BACKGROUND

Delirium is underdiagnosed in clinical practice and is not routinely coded for billing. Manual chart review can be used to identify the occurrence of delirium; however, it is labor-intensive and impractical for large-scale studies. Natural language processing (NLP) has the capability to process raw text in electronic health records (EHRs) and determine the meaning of the information. We developed and validated NLP algorithms to automatically identify the occurrence of delirium from EHRs.

METHODS

This study used a randomly selected cohort from the population-based Mayo Clinic Biobank (N = 300, age ≥65). We adopted the standardized evidence-based framework confusion assessment method (CAM) to develop and evaluate NLP algorithms to identify the occurrence of delirium using clinical notes in EHRs. Two NLP algorithms were developed based on CAM criteria: one based on the original CAM (NLP-CAM; delirium vs no delirium) and another based on our modified CAM (NLP-mCAM; definite, possible, and no delirium). The sensitivity, specificity, and accuracy were used for concordance in delirium status between NLP algorithms and manual chart review as the gold standard. The prevalence of delirium cases was examined using International Classification of Diseases, 9th Revision (ICD-9), NLP-CAM, and NLP-mCAM.

RESULTS

NLP-CAM demonstrated a sensitivity, specificity, and accuracy of 0.919, 1.000, and 0.967, respectively. NLP-mCAM demonstrated sensitivity, specificity, and accuracy of 0.827, 0.913, and 0.827, respectively. The prevalence analysis of delirium showed that the NLP-CAM algorithm identified 12 651 (9.4%) delirium patients, the NLP-mCAM algorithm identified 20 611 (15.3%) definite delirium cases, and 10 762 (8.0%) possible cases.

CONCLUSIONS

NLP algorithms based on the standardized evidence-based CAM framework demonstrated high performance in delineating delirium status in an expeditious and cost-effective manner.

Sauver JL, Larson NB. Identification of sex-specific genetic associations in response to opioid analgesics in a White, non-Hispanic cohort from Southeast Minnesota

The study of sex-specific genetic associations with opioid response may improve the understanding of inter-individual variability in pain treatments. We investigated sex-specific associations between genetic variation and opioid response. We identified participants in the RIGHT Study prescribed codeine, tramadol, hydrocodone, and oxycodone between 01/01/2005 and 12/31/2017. Prescriptions were collapsed into codeine/tramadol and hydrocodone/oxycodone. Outcomes included poor pain control and adverse reactions within six weeks after prescription date. We performed gene-level and single-variant association analyses stratified by sex. We included 7169 non-Hispanic white participants and a total of 1940 common and low-frequency variants (MAF > 0.01). Common variants in MACROD2 (rs76026520), CYP1B1 (rs1056837, rs1056836), and CYP2D6 (rs35742686) were associated with outcomes. At the gene level, FAAH, SCN1A, and TYMS had associations for men and women, and NAT2, CYP3A4, CYP1A2, and SLC22A2 had associations for men only. Our findings highlight the importance of considering sex in association studies on opioid response.

CELSR1 Risk Alleles in Familial Bicuspid Aortic Valve and Hypoplastic Left Heart Syndrome

BACKGROUND

Whole-genome sequencing in families enables deciphering of congenital heart disease causes. A shared genetic basis for familial bicuspid aortic valve (BAV) and hypoplastic left heart syndrome (HLHS) was postulated.

METHODS

Whole-genome sequencing was performed in affected members of 6 multiplex BAV families, an HLHS cohort of 197 probands and 546 relatives, and 813 controls. Data were filtered for rare, predicted-damaging variants that cosegregated with familial BAV and disrupted genes associated with congenital heart disease in humans and mice. Candidate genes were further prioritized by rare variant burden testing in HLHS cases versus controls. Modifier variants in HLHS proband-parent trios were sought to account for the severe developmental phenotype.

RESULTS

In 5 BAV families, missense variants in 6 ontologically diverse genes for structural (SPTBN1, PAXIP1, and FBLN1) and signaling (CELSR1, PLXND1, and NOS3) proteins fulfilled filtering metrics. CELSR1, encoding cadherin epidermal growth factor laminin G seven-pass G-type receptor, was identified as a candidate gene in 2 families and was the only gene demonstrating rare variant enrichment in HLHS probands (P=0.003575). HLHS-associated CELSR1 variants included 16 missense, one splice site, and 3 noncoding variants predicted to disrupt canonical transcription factor binding sites, most of which were inherited from a parent without congenital heart disease. Filtering whole-genome sequencing data for rare, predicted-damaging variants inherited from the other parent revealed 2 cases of CELSR1 compound heterozygosity, one case of CELSR1-CELSR3 synergistic heterozygosity, and 4 cases of CELSR1-MYO15A digenic heterozygosity.

CONCLUSIONS

CELSR1 is a susceptibility gene for familial BAV and HLHS, further implicating planar cell polarity pathway perturbation in congenital heart disease.

Polygenic risk score and risk of monoclonal B-cell lymphocytosis in caucasians and risk of chronic lymphocytic leukemia (CLL) in African Americans

Monoclonal B-cell lymphocytosis (MBL) is a precursor to CLL. Other than age, sex, and CLL family-history, little is known about factors associated with MBL risk. A polygenic-risk-score (PRS) of 41 CLL-susceptibility variants has been found to be associated with CLL risk among individuals of European-ancestry(EA). Here, we evaluate these variants, the PRS, and environmental factors for MBL risk. We also evaluate these variants and the CLL-PRS among African-American (AA) and EA-CLL cases and controls. Our study included 560 EA MBLs, 869 CLLs (696 EA/173 AA), and 2866 controls (2631 EA/235 AA). We used logistic regression, adjusting for age and sex, to estimate odds ratios (OR) and 95% confidence intervals within each race. We found significant associations with MBL risk among 21 of 41 variants and with the CLL-PRS (OR = 1.86, P = 1.9 × 10^-29, c-statistic = 0.72). Little evidence of any association between MBL risk and environmental factors was observed. We observed significant associations of the CLL-PRS with EA-CLL risk (OR = 2.53, P = 4.0 × 10^-63, c-statistic = 0.77) and AA-CLL risk (OR = 1.76, P = 5.1 × 10^-5, c-statistic = 0.62). Inherited genetic factors and not environmental are associated with MBL risk. In particular, the CLL-PRS is a strong predictor for both risk of MBL and EA-CLL, but less so for AA-CLL supporting the need for further work in this population.

The PSC scientific community resource: an asset for multi-omics interrogation of primary sclerosing cholangitis

Background

Primary sclerosing cholangitis (PSC) is a rare, chronic cholestatic liver disease that often progresses to end-stage liver disease and/or the development of hepatobiliary neoplasia. Lack of prognostic tools and treatment options for PSC is driven in part by our poor understanding of its pathogenesis, which is thought to be complex, the interaction of genetic variants, environmental influences and biological response throughout the course of disease. The PSC Scientific Community Resource (PSC-SCR) seeks to overcome previous shortcomings by facilitating novel research in PSC with the ultimate goals of individualizing patient care and improving patient outcomes.

Methods

PSC patients who receive their health care at Mayo Clinic or a collaborating site are identified by chart review and invited in person or by mail to participate. Non-Mayo patients are offered enrollment if they provide sufficient access to their medical records to evaluate inclusion/exclusion criteria. Controls without liver disease are identified with assistance of the Mayo Clinic Biobank. Participant consent is obtained at the beginning of the recruitment process by mail-in, electronic or face-to-face protocols. Clinical data is extracted from the medical record by qualified physicians and entered in a custom designed database. Participants fill out a custom-designed, comprehensive questionnaire, which collects scientifically relevant demographic and clinical information. Biospecimens are collected using mail-in kits thar are returned via overnight carrier service and processed by the biospecimen accessioning and processing facility at Mayo Clinic, which coordinates sample transfers and provides required sample preparation services. The resource is currently being utilized to perform omics-scale projects investigating the exposome, metabolome, methylome, immunome and microbiome in PSC. Datasets and residual biospecimens will be shared with researchers proposing scientifically sound PSC-focused research with approval of the appropriate review boards.

Discussion

Patient-based studies leveraging the latest technologies for targeted and wide-scale interrogation of multiple omics layers offer promise to accelerate PSC research through discovery of unappreciated aspects of disease pathogenesis. However, the rarity of PSC severely limits such studies. Here we describe our effort to overcome this limitation, the PSC-SCR, a repository of patient biospecimens coupled with clinical and omics data for use by the broader PSC research community.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12876-021-01930-2.

A Survey on Recent Named Entity Recognition and Relationship Extraction Techniques on Clinical Texts

Significant growth in Electronic Health Records (EHR) over the last decade has provided an abundance of clinical text that is mostly unstructured and untapped. This huge amount of clinical text data has motivated the development of new information extraction and text mining techniques. Named Entity Recognition (NER) and Relationship Extraction (RE) are key components of information extraction tasks in the clinical domain. In this paper, we highlight the present status of clinical NER and RE techniques in detail by discussing the existing proposed NLP models for the two tasks and their performances and discuss the current challenges. Our comprehensive survey on clinical NER and RE encompass current challenges, state-of-the-art practices, and future directions in information extraction from clinical text. This is the first attempt to discuss both of these interrelated topics together in the clinical context. We identified many research articles published based on different approaches and looked at applications of these tasks. We also discuss the evaluation metrics that are used in the literature to measure the effectiveness of the two these NLP methods and future research directions.

Penetrance and outcomes at 1-year following return of actionable variants identified by genome sequencing

PURPOSE

We estimated penetrance of actionable genetic variants and assessed near-term outcomes following return of results (RoR).

METHODS

Participants (n = 2,535) with hypercholesterolemia and/or colon polyps underwent targeted sequencing of 68 genes and 14 single-nucleotide variants. Penetrance was estimated based on presence of relevant traits in the electronic health record (EHR). Outcomes occurring within 1-year of RoR were ascertained by EHR review. Analyses were stratified by tier 1 and non-tier 1 disorders.

RESULTS

Actionable findings were present in 122 individuals and results were disclosed to 98. The average penetrance for tier 1 disorder variants (67%; n = 58 individuals) was higher than in non-tier 1 variants (46.5%; n = 58 individuals). After excluding 45 individuals (decedents, nonresponders, known genetic diagnoses, mosaicism), ≥1 outcomes were noted in 83% of 77 participants following RoR; 78% had a process outcome (referral to a specialist, new testing, surveillance initiated); 68% had an intermediate outcome (new test finding or diagnosis); 19% had a clinical outcome (therapy modified, risk reduction surgery). Risk reduction surgery occurred more often in participants with tier 1 than those with non-tier 1 variants.

CONCLUSION

Relevant phenotypic traits were observed in 57% whereas a clinical outcome occurred in 19% of participants with actionable genomic variants in the year following RoR.

Sex differences in type and occurrence of adverse reactions to opioid analgesics: a retrospective cohort study

OBJECTIVES

Sex as a biological variable affects response to opioids. However, few reports describe the prevalence of specific adverse reactions to commonly prescribed opioids in men and women separately. A large cohort was used to investigate sex differences in type and occurrence of adverse reactions associated with use of codeine, tramadol, oxycodone and hydrocodone.

DESIGN

Retrospective cohort study.

SETTING

Participants in the Right Drug, Right Dose, Right Time (RIGHT) Study.

PARTICIPANTS

The medical records of 8457 participants in the RIGHT Study who received an opioid prescription between 1 January 2004 and 31 December 2017 were reviewed 61% women, 94% white, median age (Q1-Q3)=58 (47-66).

PRIMARY AND SECONDARY OUTCOME MEASURES

Adverse reactions including gastrointestinal, skin, psychiatric and nervous system issues were collected from the allergy section of each patient's medical record. Sex differences in the risk of adverse reactions due to prescribed opioids were modelled using logistic regression adjusted for age, body mass index, race and ethnicity.

RESULTS

From 8457 participants (of which 449 (5.3%) reported adverse reactions), more women (6.5%) than men (3.4%) reported adverse reactions to at least one opioid (OR (95% CI)=2.3 (1.8 to 2.8), p<0.001). Women were more likely to report adverse reactions to tramadol (OR (95% CI)=2.8 (1.8 to 4.4), p<0.001) and oxycodone (OR (95% CI)=2.2 (1.7 to 2.9), p<0.001). Women were more likely to report gastrointestinal (OR (95% CI)=3.1 (2.3 to 4.3), p<0.001), skin (OR (95% CI)=2.1 (1.4 to 3.3), p=0.001) and nervous system issues (OR (95% CI)=2.3 (1.3 to 4.2), p=0.004).

CONCLUSIONS

These findings support the importance of sex as a biological variable to be factored into pain management studies.

Longitudinal cohorts for harnessing the electronic health record for disease prediction in a US population

PURPOSE

The depth and breadth of clinical data within electronic health record (EHR) systems paired with innovative machine learning methods can be leveraged to identify novel risk factors for complex diseases. However, analysing the EHR is challenging due to complexity and quality of the data. Therefore, we developed large electronic population-based cohorts with comprehensive harmonised and processed EHR data.

PARTICIPANTS

All individuals 30 years of age or older who resided in Olmsted County, Minnesota on 1 January 2006 were identified for the discovery cohort. Algorithms to define a variety of patient characteristics were developed and validated, thus building a comprehensive risk profile for each patient. Patients are followed for incident diseases and ageing-related outcomes. Using the same methods, an independent validation cohort was assembled by identifying all individuals 30 years of age or older who resided in the largely rural 26-county area of southern Minnesota and western Wisconsin on 1 January 2013.

FINDINGS TO DATE

For the discovery cohort, 76 255 individuals (median age 49; 53% women) were identified from which a total of 9 644 221 laboratory results; 9 513 840 diagnosis codes; 10 924 291 procedure codes; 1 277 231 outpatient drug prescriptions; 966 136 heart rate measurements and 1 159 836 blood pressure (BP) measurements were retrieved during the baseline time period. The most prevalent conditions in this cohort were hyperlipidaemia, hypertension and arthritis. For the validation cohort, 333 460 individuals (median age 54; 52% women) were identified and to date, a total of 19 926 750 diagnosis codes, 10 527 444 heart rate measurements and 7 356 344 BP measurements were retrieved during baseline.

FUTURE PLANS

Using advanced machine learning approaches, these electronic cohorts will be used to identify novel sex-specific risk factors for complex diseases. These approaches will allow us to address several challenges with the use of EHR.

"Who Doesn't Like Receiving Good News?" Perspectives of Individuals Who Received Genomic Screening Results by Mail

As genomic sequencing expands to screen larger numbers of individuals, offering genetic counseling to everyone may not be possible. One approach to managing this limitation is for a genetic counselor to communicate clinically actionable results in person or by telephone, but report other results by mail. We employed this approach in a large genomic implementation study. In this paper, we describe participants' experiences receiving genomic screening results by mail. We conducted 50 semi-structured telephone interviews with individuals who received neutral genomic screening results by mail. Most participants were satisfied receiving neutral results by mail. Participants generally had a good understanding of results; however, a few participants had misunderstandings about their genomic screening results, including mistaken beliefs about their disease risk and the comprehensiveness of the test. No one reported plans to alter health behaviors, defer medical evaluations, or take other actions that might be considered medically problematic. Reporting neutral results by mail is unlikely to cause recipients distress or generate misunderstandings that may result in reduced vigilance in following recommended preventive health strategies. Nonetheless, some individuals may benefit from additional genetic counseling support to help situate their results in the context of personal concerns and illness experiences.

Natural history of monoclonal B-cell lymphocytosis among relatives in CLL families

Chronic lymphocytic lymphoma (CLL) has one of the highest familial risks among cancers. Monoclonal B-cell lymphocytosis (MBL), the precursor to CLL, has a higher prevalence (13%-18%) in families with 2 or more members with CLL compared with the general population (5%-12%). Although, the rate of progression to CLL for high-count MBLs (clonal B-cell count ≥500/µL) is ∼1% to 5%/y, no low-count MBLs have been reported to progress to date. We report the incidence and natural history of MBL in relatives from CLL families. In 310 CLL families, we screened 1045 relatives for MBL using highly sensitive flow cytometry and prospectively followed 449 of them. MBL incidence was directly age- and sex-adjusted to the 2010 US population. CLL cumulative incidence was estimated using Kaplan-Meier survival curves. At baseline, the prevalence of MBL was 22% (235/1045 relatives). After a median follow-up of 8.1 years among 449 relatives, 12 individuals progressed to CLL with a 5-year cumulative incidence of 1.8%. When considering just the 139 relatives with low-count MBL, the 5-year cumulative incidence increased to 5.7%. Finally, 264 had no MBL at baseline, of whom 60 individuals subsequently developed MBL (2 high-count and 58 low-count MBLs) with an age- and sex-adjusted incidence of 3.5% after a median of 6 years of follow-up. In a screening cohort of relatives from CLL families, we reported progression from normal-count to low-count MBL to high-count MBL to CLL, demonstrating that low-count MBL precedes progression to CLL. We estimated a 1.1% annual rate of progression from low-count MBL, which is in excess of that in the general population.

Genetic basis of hypercholesterolemia in adults

We investigated monogenic and polygenic causes of hypercholesterolemia in a population-based cohort, excluding secondary hypercholesterolemia, and using an established framework to identify pathogenic variants. We studied 1682 individuals (50.2 ± 8.6 years, 41.3% males) from southeast Minnesota with primary hypercholesterolemia (low-density lipoprotein cholesterol (LDL-C) ≥155 mg/dl in the absence of identifiable secondary causes). Familial hypercholesterolemia (FH) phenotype was defined as a Dutch Lipid Clinic Network (DLCN) score ≥6. Participants underwent sequencing of LDLR, APOB, and PCSK9, and genotyping of 12 LDL-C-associated single-nucleotide variants to construct a polygenic score (PGS) for LDL-C. The presence of a pathogenic/likely pathogenic variant was considered monogenic etiology and a PGS ≥90th percentile was considered polygenic etiology. The mean LDL-C level was 187.3 ± 32.3 mg/dl and phenotypic FH was present in 8.4% of the cohort. An identifiable genetic etiology was present in 17.1% individuals (monogenic in 1.5% and polygenic in 15.6%). Phenotypic and genetic FH showed poor overlap. Only 26% of those who met the clinical criteria of FH had an identifiable genetic etiology and of those with an identifiable genetic etiology only 12.9% met clinical criteria for FH. Genetic factors explained 7.4% of the variance in LDL-C. In conclusion, in adults with primary hypercholesterolemia, 17.1% had an identifiable genetic etiology and the overlap between phenotypic and genetic FH was modest.

Adult diffuse glioma GWAS by molecular subtype identifies variants in D2HGDH and FAM20C

BACKGROUND

Twenty-five germline variants are associated with adult diffuse glioma, and some of these variants have been shown to be associated with particular subtypes of glioma. We hypothesized that additional germline variants could be identified if a genome-wide association study (GWAS) were performed by molecular subtype.

METHODS

A total of 1320 glioma cases and 1889 controls were used in the discovery set and 799 glioma cases and 808 controls in the validation set. Glioma cases were classified into molecular subtypes based on combinations of isocitrate dehydrogenase (IDH) mutation, telomerase reverse transcriptase (TERT) promoter mutation, and 1p/19q codeletion. Logistic regression was applied to the discovery and validation sets to test for associations of variants with each of the subtypes. A meta-analysis was subsequently performed using a genome-wide P-value threshold of 5 × 10-8.

RESULTS

Nine variants in or near D-2-hydroxyglutarate dehydrogenase (D2HGDH) on chromosome 2 were genome-wide significant in IDH-mutated glioma (most significant was rs5839764, meta P = 2.82 × 10-10). Further stratifying by 1p/19q codeletion status, one variant in D2HGDH was genome-wide significant in IDH-mutated non-codeleted glioma (rs1106639, meta P = 4.96 × 10-8). Further stratifying by TERT mutation, one variant near FAM20C (family with sequence similarity 20, member C) on chromosome 7 was genome-wide significant in gliomas that have IDH mutation, TERT mutation, and 1p/19q codeletion (rs111976262, meta P = 9.56 × 10-9). Thirty-six variants in or near GMEB2 on chromosome 20 near regulator of telomere elongation helicase 1 (RTEL1) were genome-wide significant in IDH wild-type glioma (most significant was rs4809313, meta P = 2.60 × 10-10).

CONCLUSIONS

Performing a GWAS by molecular subtype identified 2 new regions and a candidate independent region near RTEL1, which were associated with specific glioma molecular subtypes.

Investigating the impact of disease and health record duration on the eMERGE algorithm for rheumatoid arthritis

OBJECTIVE

The study sought to determine the dependence of the Electronic Medical Records and Genomics (eMERGE) rheumatoid arthritis (RA) algorithm on both RA and electronic health record (EHR) duration.

MATERIALS AND METHODS

Using a population-based cohort from the Mayo Clinic Biobank, we identified 497 patients with at least 1 RA diagnosis code. RA case status was manually determined using validated criteria for RA. RA duration was defined as time from first RA code to the index date of biobank enrollment. To simulate EHR duration, various years of EHR lookback were applied, starting at the index date and going backward. Model performance was determined by sensitivity, specificity, positive predictive value, negative predictive value, and area under the curve (AUC).

RESULTS

The eMERGE algorithm performed well in this cohort, with overall sensitivity 53%, specificity 99%, positive predictive value 97%, negative predictive value 74%, and AUC 76%. Among patients with RA duration <2 years, sensitivity and AUC were only 9% and 54%, respectively, but increased to 71% and 85% among patients with RA duration >10 years. Longer EHR lookback also improved model performance up to a threshold of 10 years, in which sensitivity reached 52% and AUC 75%. However, optimal EHR lookback varied by RA duration; an EHR lookback of 3 years was best able to identify recently diagnosed RA cases.

CONCLUSIONS

eMERGE algorithm performance improves with longer RA duration as well as EHR duration up to 10 years, though shorter EHR lookback can improve identification of recently diagnosed RA cases.

Body mass index and blood pressure in bipolar patients: Target cardiometabolic markers for clinical practice

OBJECTIVE

To evaluate the association between cardiometabolic markers and bipolar disorder (BD), examining the impact of sex and cardiometabolic medication use, from a large case-control biorepository of more than 1300 participants.

PATIENTS AND METHODS

Recruited from July 2009 through September 2017, cardiometabolic markers were harvested from electronic health records (EHR) of participants (n=661) from the Mayo Clinic Individualized Medicine Biobank for Bipolar Disorder and Mayo Clinic Biobank age-sex-matched controls (n=706). Markers were compared between cases and controls using logistic regression, stratified by sex, adjusting for cardiometabolic medications and current smoking status. We studied the effect of psychotropics in case-only analyses.

RESULTS

The mean age of the sample was 52.5 ± 11.6 years and 55% were female. BD patients had higher rates of smoking, but lower utilization of lipid-lowering medication compared with controls. After adjustment, BD was associated with obesity [Odds ratio (CI) 1.62 (1.22-2.15)], elevated systolic blood pressure (SBP) [2.18 (1.55-3.06)] and elevated triglycerides [1.58 (1.13-2.2)]. When stratified by sex, obesity [1.8 (1.23-2.66)] and systolic blood pressure [2.32 (1.46-3.7)] were associated with BD females compared to female controls; however, only systolic blood pressure [2.04 (1.23-3.42)] was associated with male bipolars compared to male controls. Psychotropics were marginally associated with mean BMI, abnormal triglycerides, and HbA1c.

LIMITATIONS

EHR cross-sectional data CONCLUSION: To our knowledge, this is the largest case controlled study to date to explore the association between cardiometabolic markers and bipolar disorder adjusting for utilization of cardiometabolic medication. Identification of significant, non-laboratory based cardiometabolic markers that are associated with increased risk of major cardiovascular adverse events in patients with bipolar disorder, underscores, both the utility and importance of risk monitoring that can be easily done in community mental health centers.

Family History of Rheumatic, Autoimmune, and Nonautoimmune Diseases and Risk of Rheumatoid Arthritis

OBJECTIVE

Since comorbidities such as autoimmune diseases may be associated with rheumatoid arthritis (RA) risk, we hypothesized that a family history of these other conditions might also predict RA. Therefore, we aimed to determine the association between family history of 79 comorbidities and RA.

METHODS

This case-control study identified 821 cases of RA in the Mayo Clinic Biobank (positive predictive value 95%) and matched 3 controls to each case based on age, sex, recruitment year, and location. Patients self reported family history and characteristics (adjusted). Logistic regression was used to estimate odds ratios (ORs) and 95% confidence intervals (95% CIs) for RA risk according to the presence of family history for each comorbidity, adjusted for body mass index, race, and smoking.

RESULTS

Family history of several conditions was associated with developing RA, including rheumatic autoimmune diseases (ORadj 1.89 [95% CI 1.41-2.52]), pulmonary fibrosis (ORadj 2.12 [95% CI 1.16-3.80]), inflammatory bowel disease (ORadj 1.45 [95% CI 1.05-1.98]), hyper/hypothyroidism (ORadj 1.34 [95% CI 1.10-1.63]), and obstructive sleep apnea (ORadj 1.28 [95% CI 1.05-1.55]). Parkinson's disease and type 2 diabetes mellitus were associated with a statistically decreased risk of RA that did not reach the prespecified significance threshold of P < 0.01 (ORadj 0.70 [95% CI 0.49-0.98] and ORadj 0.81 [95% CI 0.67-0.97], respectively). Analyses among 143 cases of incident RA were similar and also suggested an association with a family history of autism (OR 10.5 [95% CI 2.51-71.3]).

CONCLUSION

Family history of several autoimmune and nonautoimmune comorbidities was associated with increased risk of RA, providing an opportunity to identify novel populations at risk for RA.

Increasing access to individualized medicine: a matched-cohort study examining Latino participant experiences of genomic screening

Purpose:

Multiple efforts are underway to increase the inclusion of racial minority participants in genomic research and new forms of individualized medicine. These efforts should include studies that characterize how individuals from minority communities experience genomic medicine in diverse healthcare settings and how they integrate genetic knowledge into their understandings of healthcare needs.

Methods:

As part of a large, multisite genomic sequencing study, we surveyed individuals to assess their decision to pursue genomic risk evaluation. Participants included Latino patients recruited at Mountain Park Health Center, a Federally Qualified Health Center in Phoenix, AZ, and non-Latino patients recruited at a large academic medical center (Mayo Clinic in Rochester, MN). Both groups agreed to receive individualized genomic risk assessments.

Results:

Comparisons between cohorts showed that Latino respondents had lower levels of decisional conflict about pursuing genomic screening but generally scored lower on genetic knowledge. Latino respondents were also more likely to have concerns about the misuse of genomic information, despite both groups having similar views about the value of genomic risk evaluation.

Conclusion:

Our results highlight the importance of evaluating sociocultural factors that influence minority patient engagement with genomic medicine in diverse healthcare settings.

A Cohort Study Examining the Interaction of Alcohol Consumption and Obesity in Hepatic Steatosis and Mortality

OBJECTIVE

To study the interaction of alcohol consumption with body mass index (BMI) in the development of hepatic steatosis and mortality.

PARTICIPANTS AND METHODS

We conducted a retrospective cohort study of 18,506 participants without fatty liver disease or cirrhosis at enrollment in the Mayo Clinic Biobank from April 9, 2009, through March 31, 2016. Participants were classified by self-reported alcohol consumption status (nondrinkers, moderate drinkers [0 to 2 drinks per day], and heavy drinkers [>2 drinks per day]). The primary outcome of interest was the incidence of hepatic steatosis, identified by International Classification of Diseases, Ninth Revision code and confirmed with imaging. The secondary outcome of interest was all-cause mortality. Multivariate Cox regression analysis determined the impact of alcohol consumption stratified by BMI on outcomes compared with nondrinkers.

RESULTS

The cohort (mean ± SD age, 55.8±16.9 years; 63.8% female; mean ± SD BMI, 28.8±6.1 kg/m²) of 18,506 participants included 3657 (19.8%) nondrinkers, 14,236 (76.9%) moderate drinkers, and 613 (3.3%) heavy drinkers at enrollment. After a median follow-up of 5.8 years (interquartile range, 3.8 to 7.2 years), 684 participants had development of hepatic steatosis and 968 died. In moderate drinkers, the risk of hepatic steatosis development was high in the obese group (adjusted hazard ratio [AHR], 1.31; 95% CI, 1.03 to 1.67), insignificant in the overweight group (AHR, 0.86; 95% CI, 0.58 to 1.26), and decreased in the normal-BMI group (AHR, 0.48; 95% CI, 0.26 to 0.90). Heavy drinkers had an increased risk of hepatic steatosis irrespective of BMI. Moderate alcohol use was associated with decreased mortality in the normal-weight (AHR, 0.44; 95% CI, 0.34 to 0.58) and overweight (AHR, 0.70; 95% CI, 0.56 to 0.88) groups but not in the obese group (AHR, 0.80; 95% CI, 0.64 to 1.00).

CONCLUSION

In obese individuals, even moderate alcohol use is associated with the development of hepatic steatosis. Moderate alcohol consumption is associated with lower mortality in normal-BMI and overweight individuals but not in those who are obese.

Parent Perspectives Towards Genetic and Epigenetic Testing for Autism Spectrum Disorder

Examining community views on genetic/epigenetic research allows collaborative technology development. Parent perspectives toward genetic/epigenetic testing for autism spectrum disorder (ASD) are not well-studied. Parents of children with ASD (n = 131), non-ASD developmental delay (n = 39), and typical development (n = 74) completed surveys assessing genetic/epigenetic knowledge, genetic/epigenetic concerns, motives for research participation, and attitudes/preferences toward ASD testing. Most parents (96%) were interested in saliva-based molecular testing for ASD. Some had concerns about privacy (14%) and insurance-status (10%). None (0%) doubted scientific evidence behind genetic/epigenetic testing. Most reported familiarity with genetics (88%), but few understood differences from epigenetics (19%). Child developmental status impacted insurance concerns (p = 0.01). There is broad parent interest in a genetic/epigenetic test for ASD. It will be crucial to carefully consider and address bioethical issues surrounding this sensitive topic while developing such technology.

Implementation of a Cohort Retrieval System for Clinical Data Repositories Using the Observational Medical Outcomes Partnership Common Data Model: Proof-of-Concept System Validation

BACKGROUND

Widespread adoption of electronic health records has enabled the secondary use of electronic health record data for clinical research and health care delivery. Natural language processing techniques have shown promise in their capability to extract the information embedded in unstructured clinical data, and information retrieval techniques provide flexible and scalable solutions that can augment natural language processing systems for retrieving and ranking relevant records.

OBJECTIVE

In this paper, we present the implementation of a cohort retrieval system that can execute textual cohort selection queries on both structured data and unstructured text-Cohort Retrieval Enhanced by Analysis of Text from Electronic Health Records (CREATE).

METHODS

CREATE is a proof-of-concept system that leverages a combination of structured queries and information retrieval techniques on natural language processing results to improve cohort retrieval performance using the Observational Medical Outcomes Partnership Common Data Model to enhance model portability. The natural language processing component was used to extract common data model concepts from textual queries. We designed a hierarchical index to support the common data model concept search utilizing information retrieval techniques and frameworks.

RESULTS

Our case study on 5 cohort identification queries, evaluated using the precision at 5 information retrieval metric at both the patient-level and document-level, demonstrates that CREATE achieves a mean precision at 5 of 0.90, which outperforms systems using only structured data or only unstructured text with mean precision at 5 values of 0.54 and 0.74, respectively.

CONCLUSIONS

The implementation and evaluation of Mayo Clinic Biobank data demonstrated that CREATE outperforms cohort retrieval systems that only use one of either structured data or unstructured text in complex textual cohort queries.

Patient-specific genomics and cross-species functional analysis implicate LRP2 in hypoplastic left heart syndrome

Congenital heart diseases (CHDs), including hypoplastic left heart syndrome (HLHS), are genetically complex and poorly understood. Here, a multidisciplinary platform was established to functionally evaluate novel CHD gene candidates, based on whole-genome and iPSC RNA sequencing of a HLHS family-trio. Filtering for rare variants and altered expression in proband iPSCs prioritized 10 candidates. siRNA/RNAi-mediated knockdown in healthy human iPSC-derived cardiomyocytes (hiPSC-CM) and in developing Drosophila and zebrafish hearts revealed that LDL receptor-related protein LRP2 is required for cardiomyocyte proliferation and differentiation. Consistent with hypoplastic heart defects, compared to patents the proband's iPSC-CMs exhibited reduced proliferation. Interestingly, rare, predicted-damaging LRP2 variants were enriched in a HLHS cohort; however, understanding their contribution to HLHS requires further investigation. Collectively, we have established a multi-species high-throughput platform to rapidly evaluate candidate genes and their interactions during heart development, which are crucial first steps toward deciphering oligogenic underpinnings of CHDs, including hypoplastic left hearts.

Association of Genetic Variants at TRPC6 With Chemotherapy-Related Heart Failure

Background: Our previous GWAS identified genetic variants at six novel loci that were associated with a decline in left ventricular ejection fraction (LVEF), p < 1 × 10⁻⁵ in 1,191 early breast cancer patients from the N9831 clinical trial of chemotherapy plus trastuzumab. In this study we sought replication of these loci.

Methods: We tested the top loci from the GWAS for association with chemotherapy-related heart failure (CRHF) using 26 CRHF cases from N9831 and 984 patients from the Mayo Clinic Biobank which included CRHF cases (N = 12) and control groups of patients treated with anthracycline +/– trastuzumab without HF (N = 282) and patients with HF that were never treated with anthracycline or trastuzumab (N = 690). We further examined associated loci in the context of gene expression and rare coding variants using a TWAS approach in heart left ventricle and Sanger sequencing, respectively. Doxorubicin-induced apoptosis and cardiomyopathy was modeled in human iPSC-derived cardiomyocytes and endothelial cells and a mouse model, respectively, that were pre-treated with GsMTx-4, an inhibitor of TRPC6.

Results:TRPC6 5′ flanking variant rs57242572-T was significantly more frequent in cases compared to controls, p = 0.031, and rs61918162-T showed a trend for association, p = 0.065. The rs61918162 T-allele was associated with higher TRPC6 expression in the heart left ventricle. We identified a single TRPC6 rare missense variant (rs767086724, N338S, prevalence 0.0025% in GnomAD) in one of 38 patients (2.6%) with CRHF. Pre-treatment of cardiomyocytes and endothelial cells with GsMTx4 significantly reduced doxorubicin-induced apoptosis. Similarly, mice treated with GsMTx4 had significantly improved doxorubicin-induced cardiac dysfunction.

Conclusions: Genetic variants that are associated with increased TRPC6 expression in the heart and rare TRPC6 missense variants may be clinically useful as risk factors for CRHF. GsMTx-4 may be a cardioprotective agent in patients with TRPC6 risk variants. Replication of the genetic associations in larger well-characterized samples and functional studies are required.

Identification of Susceptibility Loci for Spontaneous Coronary Artery Dissection

Importance

Spontaneous coronary artery dissection (SCAD), an idiopathic disorder that predominantly affects young to middle-aged women, has emerged as an important cause of acute coronary syndrome, myocardial infarction, and sudden cardiac death.

Objective

To identify common single-nucleotide variants (SNVs) associated with SCAD susceptibility.

Design, Setting, and Participants

This single-center genome-wide association study examined approximately 5 million genotyped and imputed SNVs and subsequent SNV-targeted replication analysis results in individuals enrolled in the Mayo Clinic SCAD registry from August 30, 2011, to August 2, 2018. Data analysis was performed from June 21, 2017, to December 30, 2019.

Main Outcomes and Measures

Genetic loci and positional candidate genes associated with SCAD.

Results

This study included 484 white women with SCAD (mean [SD] age, 46.6 [9.2] years) and 1477 white female controls in the discovery cohort (mean [SD] age, 64.0 [14.5] years) and 183 white women with SCAD (mean [SD] age, 47.1 [9.9] years) and 340 white female controls in the replication cohort (mean [SD] age, 51.0 [15.3] years). Associations with SCAD risk reached genome-wide significance at 3 loci (1q21.3 [OR, 1.78; 95% CI, 1.51-2.09; P = 2.63 × 10-12], 6p24.1 [OR, 1.77; 95% CI, 1.51-2.09; P = 7.09 × 10-12], and 12q13.3 [OR, 1.67; 95% CI, 1.42-1.97; P = 3.62 × 10-10]), and 7 loci had evidence suggestive of an association (1q24.2 [OR, 2.10; 95% CI, 1.58-2.79; P = 2.88 × 10-7], 3q22.3 [OR, 1.47; 95% CI, 1.26-1.71; P = 6.65 × 10-7], 4q34.3 [OR, 1.84; 95% CI, 1.44-2.35; P = 9.80 × 10-7], 8q24.3 [OR, 2.57; 95% CI, 1.76-3.75; P = 9.65 × 10-7], 15q21.1 [OR, 1.75; 95% CI, 1.40-2.18; P = 7.23 × 10-7], 16q24.1 [OR, 1.91; 95% CI, 1.49-2.44; P = 2.56 × 10-7], and 21q22.11 [OR, 2.11; 95% CI, 1.59-2.82; P = 3.12 × 10-7]) after adjusting for the top 5 principal components. Associations were validated for 5 of the 10 risk alleles in the replication cohort. In a meta-analysis of the discovery and replication cohorts, associations for the 5 SNVs were significant, with relatively large effect sizes (1q21.3 [OR, 1.77; 95% CI, 1.54-2.03; P = 3.26 × 10-16], 6p24.1 [OR, 1.71; 95% CI, 1.49-1.97; P = 4.59 × 10-14], 12q13.3 [OR, 1.69; 95% CI, 1.47-1.94; P = 1.42 × 10-13], 15q21.1 [OR, 1.79; 95% CI, 1.48-2.17; P = 2.12 × 10-9], and 21q22.11 [OR, 2.18; 95% CI, 1.70-2.81; P = 1.09 × 10-9]). Each index SNV was within or near a gene highly expressed in arterial tissue and previously linked to SCAD (PHACTR1) and/or other vascular disorders (LRP1, LINC00310, and FBN1).

Conclusions and Relevance

This study revealed 5 replicated risk loci and positional candidate genes for SCAD, most of which are associated with extracoronary arteriopathies. Moreover, the alternate alleles of 3 SNVs have been previously associated with atherosclerotic coronary artery disease, further implicating allelic susceptibility to coronary artery atherosclerosis vs dissection.

The reported impact of public involvement in biobanks: A scoping review

BACKGROUND

Biobanks increasingly employ public involvement and engagement strategies, though few studies have explored their impact. This review aims to (a) investigate how the impact of public involvement in biobanks is reported and conceptualized by study authors; in order to (b) suggest how the research community might re-conceptualize the impact of public involvement in biobanks.

METHODS

A systematic literature search of three electronic databases and the INVOLVE Evidence Library in January 2019. Studies commenting on the impact of public involvement in a biobank were included, and a narrative review was conducted.

RESULTS AND DISCUSSION

Forty-one studies covering thirty-one biobanks were included, with varying degrees of public involvement. Impact was categorized according to where it was seen: 'the biobank', 'people involved' and 'the wider research community'. Most studies reported involvement in a 'functional' way, in relation to improved rates of participation in the biobank. Broader forms of impact were reported but were vaguely defined and measured. This review highlights a lack of clarity of purpose and varied researcher conceptualizations of involvement. We pose three areas for further research and consideration by biobank researchers and public involvement practitioners.

CONCLUSIONS

Functional approaches to public involvement in biobanking limit impact. This conceptualization of involvement emerges from an entrenched technical understanding that ignores its political nature, complicated by long-standing disagreement about the values of public involvement. This study urges a re-imagination of impact, re-conceptualized as a two-way learning process. More support will help researchers and members of the public to undergo such reflective exercises.

Next-Generation Sequencing of CYP2C19 in Stent Thrombosis: Implications for Clopidogrel Pharmacogenomics

PURPOSE

Describe CYP2C19 sequencing results in the largest series of clopidogrel-treated cases with stent thrombosis (ST), the closest clinical phenotype to clopidogrel resistance. Evaluate the impact of CYP2C19 genetic variation detected by next-generation sequencing (NGS) with comprehensive annotation and functional studies.

METHODS

Seventy ST cases on clopidogrel identified from the PLATO trial (n = 58) and Mayo Clinic biorepository (n = 12) were matched 1:1 with controls for age, race, sex, diabetes mellitus, presentation, and stent type. NGS was performed to cover the entire CYP2C19 gene. Assessment of exonic variants involved measuring in vitro protein expression levels. Intronic variants were evaluated for potential splicing motif variations.

RESULTS

Poor metabolizers (n = 4) and rare CYP2C19*8, CYP2C19*15, and CYP2C19*11 alleles were identified only in ST cases. CYP2C19*17 heterozygote carriers were observed more frequently in cases (n = 29) than controls (n = 18). Functional studies of CYP2C19 exonic variants (n = 11) revealed 3 cases and only 1 control carrying a deleterious variant as determined by in vitro protein expression studies. Greater intronic variation unique to ST cases (n = 169) compared with controls (n = 84) was observed with predictions revealing 13 allele candidates that may lead to a potential disruption of splicing and a loss-of-function effect of CYP2C19 in ST cases.

CONCLUSION

NGS detected CYP2C19 poor metabolizers and paradoxically greater number of so-called rapid metabolizers in ST cases. Rare deleterious exonic variation occurs in 4%, and potentially disruptive intronic alleles occur in 16% of ST cases. Additional studies are required to evaluate the role of these variants in platelet aggregation and clopidogrel metabolism.

The senescence-associated secretome as an indicator of age and medical risk

Produced by senescent cells, the senescence-associated secretory phenotype (SASP) is a potential driver of age-related dysfunction. We tested whether circulating concentrations of SASP proteins reflect age and medical risk in humans. We first screened senescent endothelial cells, fibroblasts, preadipocytes, epithelial cells, and myoblasts to identify candidates for human profiling. We then tested associations between circulating SASP proteins and clinical data from individuals throughout the life span and older adults undergoing surgery for prevalent but distinct age-related diseases. A community-based sample of people aged 20-90 years (retrospective cross-sectional) was studied to test associations between circulating SASP factors and chronological age. A subset of this cohort aged 60-90 years and separate cohorts of older adults undergoing surgery for severe aortic stenosis (prospective longitudinal) or ovarian cancer (prospective case-control) were studied to assess relationships between circulating concentrations of SASP proteins and biological age (determined by the accumulation of age-related health deficits) and/or postsurgical outcomes. We showed that SASP proteins were positively associated with age, frailty, and adverse postsurgery outcomes. A panel of 7 SASP factors composed of growth differentiation factor 15 (GDF15), TNF receptor superfamily member 6 (FAS), osteopontin (OPN), TNF receptor 1 (TNFR1), ACTIVIN A, chemokine (C-C motif) ligand 3 (CCL3), and IL-15 predicted adverse events markedly better than a single SASP protein or age. Our findings suggest that the circulating SASP may serve as a clinically useful candidate biomarker of age-related health and a powerful tool for interventional human studies.

Impact of Diverse Data Sources on Computational Phenotyping

Electronic health records (EHRs) are widely adopted with a great potential to serve as a rich, integrated source of phenotype information. Computational phenotyping, which extracts phenotypes from EHR data automatically, can accelerate the adoption and utilization of phenotype-driven efforts to advance scientific discovery and improve healthcare delivery. A list of computational phenotyping algorithms has been published but data fragmentation, i.e., incomplete data within one single data source, has been raised as an inherent limitation of computational phenotyping. In this study, we investigated the impact of diverse data sources on two published computational phenotyping algorithms, rheumatoid arthritis (RA) and type 2 diabetes mellitus (T2DM), using Mayo EHRs and Rochester Epidemiology Project (REP) which links medical records from multiple health care systems. Results showed that both RA (less prevalent) and T2DM (more prevalent) case selections were markedly impacted by data fragmentation, with positive predictive value (PPV) of 91.4 and 92.4%, false-negative rate (FNR) of 26.6 and 14% in Mayo data, respectively, PPV of 97.2 and 98.3%, FNR of 5.2 and 3.3% in REP. T2DM controls also contain biases, with PPV of 91.2% and FNR of 1.2% for Mayo. We further elaborated underlying reasons impacting the performance.

The urgency of utilizing COVID-19 biospecimens for research in the heart of the global pandemic

The outbreak of the novel coronavirus disease 2019 (COVID-19) and consequent social distancing practices have disrupted essential clinical research functions worldwide. Ironically, this coincides with an immediate need for research to comprehend the biology of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) and the pathology of COVID-19. As the global crisis has already led to over 15,000 deaths out of 175,000 confirmed cases in New York City and Nassau County, NY alone, it is increasingly urgent to collect patient biospecimens linked to active clinical follow up. However, building a COVID-19 biorepository amidst the active pandemic is a complex and delicate task. To help facilitate rapid, robust, and regulated research on this novel virus, we report on the successful model implemented by New York University Langone Health (NYULH) within days of outbreak in the most challenging hot spot of infection globally. Using an amended institutional biobanking protocol, these efforts led to accrual of 11,120 patients presenting for SARS-CoV-2 testing, 4267 (38.4%) of whom tested positive for COVID-19. The recently reported genomic characterization of SARS-CoV-2 in the New York City Region, which is a crucial development in tracing sources of infection and asymptomatic spread of the novel virus, is the first outcome of this effort. While this growing resource actively supports studies of the New York outbreak in real time, a worldwide effort is necessary to build a collective arsenal of research tools to deal with the global crisis now, and to exploit the virus's biology for translational innovation that outlasts humanity's current dilemma.

Inferring multimodal latent topics from electronic health records

Electronic health records (EHR) are rich heterogeneous collections of patient health information, whose broad adoption provides clinicians and researchers unprecedented opportunities for health informatics, disease-risk prediction, actionable clinical recommendations, and precision medicine. However, EHRs present several modeling challenges, including highly sparse data matrices, noisy irregular clinical notes, arbitrary biases in billing code assignment, diagnosis-driven lab tests, and heterogeneous data types. To address these challenges, we present MixEHR, a multi-view Bayesian topic model. We demonstrate MixEHR on MIMIC-III, Mayo Clinic Bipolar Disorder, and Quebec Congenital Heart Disease EHR datasets. Qualitatively, MixEHR disease topics reveal meaningful combinations of clinical features across heterogeneous data types. Quantitatively, we observe superior prediction accuracy of diagnostic codes and lab test imputations compared to the state-of-art methods. We leverage the inferred patient topic mixtures to classify target diseases and predict mortality of patients in critical conditions. In all comparison, MixEHR confers competitive performance and reveals meaningful disease-related topics.