DETECT: Feature extraction method for disease trajectory modeling in electronic health records.

Modeling with longitudinal electronic health record (EHR) data proves challenging given the high dimensionality, redundancy, and noise captured in EHR. In order to improve precision medicine strategies and identify predictors of disease risk in advance, evaluating meaningful patient disease trajectories is essential. In this study, we develop the algorithm DiseasE Trajectory fEature extraCTion (DETECT) for feature extraction and trajectory generation in high-throughput temporal EHR data. This algorithm can 1) simulate longitudinal individual-level EHR data, specified to user parameters of scale, complexity, and noise and 2) use a convergent relative risk framework to test intermediate codes occurring between specified index code(s) and outcome code(s) to determine if they are predictive features of the outcome. Temporal range can be specified to investigate predictors occurring during a specific period of time prior to onset of the outcome. We benchmarked our method on simulated data and generated real-world disease trajectories using DETECT in a cohort of 145,575 individuals diagnosed with hypertension in Penn Medicine EHR for severe cardiometabolic outcomes.

Incremental prognostic value of visually estimated coronary artery calcium in patients undergoing positron emission tomography imaging.

OBJECTIVE: Visually estimated coronary artery calcium (VECAC) from chest CT or attenuation correction (AC)/CT obtained during positron emission tomography (PET)-myocardial perfusion imaging (MPI) is feasible. Our aim was to determine the prognostic value of VECAC beyond conventional risk factors and PET imaging parameters, including coronary flow reserve (CFR). METHODS: We analysed 608 patients without known coronary artery disease who underwent PET-MPI between 2012 and 2016 and had AC/CT and/or chest CT images. We used Cox regression to estimate the association of VECAC categories (≤10, 11-400, >400 Agatston units (AU)) with the primary outcome of all-cause death, acute coronary syndrome or stroke (mean follow-up 4.3±1.8 years). C-statistics assessed the relationship between PET parameters and VECAC with the primary outcome. RESULTS: Mean age was 58±11 years, 65% were women and 67% were black. VECAC ≤10, 11-400 and >400 AU was observed in 68%, 12% and 20% of subjects, respectively. Compared with VECAC ≤10, VECAC categories 11-400 (HR 2.25, 95% CI 1.24 to 4.08) and >400 AU (HR 3.05, 95% CI 1.87 to 4.98) were associated with the primary outcome after adjusting for traditional risk factors, MPI findings and CFR. Adding VECAC to a model that included PET-MPI, CFR and clinical risk factors improved the prognostic value for the primary outcomes (c-statistic 0.71 to 0.75 with VECAC, p=0.01). CONCLUSIONS: VECAC is a potent predictor of events beyond traditional risk factors and PET imaging markers, including CFR. These data further support the importance for routine VECAC implementation.

Quantification of defect contrast in microSPECT imaging of a myocardial phantom.

Ischemic heart disease remains a significant public health concern, accentuating the importance of basic research and therapeutic studies of small animals in which myocardial changes can be reproducibly detected and quantified. Few or no studies have investigated the performance of microSPECT in quantifying myocardial lesions. We utilized three versions of a multi-compartment phantom containing two left ventricular myocardial compartments (one uniform and one with a transmural 'cold' defect), a ventricular blood pool, and a background compartment, where each version had a different myocardial wall thickness (0.75, 1.0 and 1.25 mm). Each compartment was imaged separately while acquiring list-mode data. The separate compartment data were manipulated into a single data set with a known defect contrast, blood-pool and background activity. Data were processed with background-free defect-contrast values of 0 (no defect), -0.25, -0.5, -0.75, and -1.0 (all defect), three ratios of blood-pool to myocardial activity, 0 (no blood pool activity), 0.1, and 0.2 (20% of the activity in the healthy myocardial compartment), and three ratios of uniform background 0 (no background activity), 0.1 and 0.2, relative to the healthy myocardial compartment. For each wall thickness, defect contrast, blood-pool, and background activity combination, 25 list-mode noise realizations were generated and reconstructed. Volumes of interest were drawn and used to determine mean contrast recovery coefficients (CRCs) over the noise ensembles. We developed a slope-analysis procedure to estimate a single CRC over all contrast levels, with resulting CRC values (for no blood-pool and no background) of 0.848, 0.946, and 0.834 for the 0.75, 1.0, and 1.25 mm wall thicknesses, respectively. We also determined and validated a reprocessing method to calculate an ideal CRC. This work demonstrates the quantitative abilities of microSPECT for myocardial-defect imaging utilizing CRC and establishes a framework for evaluating defect-imaging capabilities in other systems.

The role of resting myocardial blood flow and myocardial blood flow reserve as a predictor of major adverse cardiovascular outcomes.

Cardiac perfusion PET is increasingly used to assess ischemia and cardiovascular risk and can also provide quantitative myocardial blood flow (MBF) and flow reserve (MBFR) values. These have been shown to be prognostic biomarkers of adverse outcomes, yet MBF and MBFR quantification remains underutilized in clinical settings. We compare MBFR to traditional cardiovascular risk factors in a large and diverse clinical population (60% African-American, 35.3% Caucasian) to rank its relative contribution to cardiovascular outcomes. Major adverse cardiovascular events (MACE), including unstable angina, non-ST and ST-elevation myocardial infarction, stroke, and death, were assessed for consecutive patients who underwent rest-dipyridamole stress 82Rb PET cardiac imaging from 2012-2015 at the Hospital of the University of Pennsylvania (n = 1283, mean follow-up 2.3 years). Resting MBF (1.1 ± 0.4 ml/min/g) was associated with adverse cardiovascular outcomes. MBFR (2.1 ± 0.8) was independently and inversely associated with MACE. Furthermore, MBFR was more strongly associated with MACE than both traditional cardiovascular risk factors and the presence of perfusion defects in regression analysis. Decision tree analysis identified MBFR as superior to established cardiovascular risk factors in predicting outcomes. Incorporating resting MBF and MBFR in CAD assessment may improve clinical decision making.

Genomic profiling of human vascular cells identifies TWIST1 as a causal gene for common vascular diseases.

Genome-wide association studies have identified multiple novel genomic loci associated with vascular diseases. Many of these loci are common non-coding variants that affect the expression of disease-relevant genes within coronary vascular cells. To identify such genes on a genome-wide level, we performed deep transcriptomic analysis of genotyped primary human coronary artery smooth muscle cells (HCASMCs) and coronary endothelial cells (HCAECs) from the same subjects, including splicing Quantitative Trait Loci (sQTL), allele-specific expression (ASE), and colocalization analyses. We identified sQTLs for TARS2, YAP1, CFDP1, and STAT6 in HCASMCs and HCAECs, and 233 ASE genes, a subset of which are also GTEx eGenes in arterial tissues. Colocalization of GWAS association signals for coronary artery disease (CAD), migraine, stroke and abdominal aortic aneurysm with GTEx eGenes in aorta, coronary artery and tibial artery discovered novel candidate risk genes for these diseases. At the CAD and stroke locus tagged by rs2107595 we demonstrate colocalization with expression of the proximal gene TWIST1. We show that disrupting the rs2107595 locus alters TWIST1 expression and that the risk allele has increased binding of the NOTCH signaling protein RBPJ. Finally, we provide data that TWIST1 expression influences vascular SMC phenotypes, including proliferation and calcification, as a potential mechanism supporting a role for TWIST1 in CAD.

Association of the V122I Hereditary Transthyretin Amyloidosis Genetic Variant With Heart Failure Among Individuals of African or Hispanic/Latino Ancestry.

Importance: Hereditary transthyretin (TTR) amyloid cardiomyopathy (hATTR-CM) due to the TTR V122I variant is an autosomal-dominant disorder that causes heart failure in elderly individuals of African ancestry. The clinical associations of carrying the variant, its effect in other African ancestry populations including Hispanic/Latino individuals, and the rates of achieving a clinical diagnosis in carriers are unknown. Objective: To assess the association between the TTR V122I variant and heart failure and identify rates of hATTR-CM diagnosis among carriers with heart failure. Design, Setting, and Participants: Cross-sectional analysis of carriers and noncarriers of TTR V122I of African ancestry aged 50 years or older enrolled in the Penn Medicine Biobank between 2008 and 2017 using electronic health record data from 1996 to 2017. Case-control study in participants of African and Hispanic/Latino ancestry with and without heart failure in the Mount Sinai BioMe Biobank enrolled between 2007 and 2015 using electronic health record data from 2007 to 2018. Exposures: TTR V122I carrier status. Main Outcomes and Measures: The primary outcome was prevalent heart failure. The rate of diagnosis with hATTR-CM among TTR V122I carriers with heart failure was measured. Results: The cross-sectional cohort included 3724 individuals of African ancestry with a median age of 64 years (interquartile range, 57-71); 1755 (47%) were male, 2896 (78%) had a diagnosis of hypertension, and 753 (20%) had a history of myocardial infarction or coronary revascularization. There were 116 TTR V122I carriers (3.1%); 1121 participants (30%) had heart failure. The case-control study consisted of 2307 individuals of African ancestry and 3663 Hispanic/Latino individuals; the median age was 73 years (interquartile range, 68-80), 2271 (38%) were male, 4709 (79%) had a diagnosis of hypertension, and 1008 (17%) had a history of myocardial infarction or coronary revascularization. There were 1376 cases of heart failure. TTR V122I was associated with higher rates of heart failure (cross-sectional cohort: n = 51/116 TTR V122I carriers [44%], n = 1070/3608 noncarriers [30%], adjusted odds ratio, 1.7 [95% CI, 1.2-2.4], P = .006; case-control study: n = 36/1376 heart failure cases [2.6%], n = 82/4594 controls [1.8%], adjusted odds ratio, 1.8 [95% CI, 1.2-2.7], P = .008). Ten of 92 TTR V122I carriers with heart failure (11%) were diagnosed as having hATTR-CM; the median time from onset of symptoms to clinical diagnosis was 3 years. Conclusions and Relevance: Among individuals of African or Hispanic/Latino ancestry enrolled in 2 academic medical center-based biobanks, the TTR V122I genetic variant was significantly associated with heart failure.

Quantification of myocardial uptake rate constants in dynamic small-animal SPECT using a cardiac phantom.

Myocardial blood flow and myocardial blood flow reserve (MBFR) measurements are often used clinically to quantify coronary microvascular function. Developing imaging-based methods to measure MBFR for research in mice would be advantageous for evaluating new treatment methods for coronary microvascular disease (CMVD), yet this is more challenging in mice than in humans. This work investigates microSPECT's quantitative capabilities of cardiac imaging by utilizing a multi-part cardiac phantom and applying a known kinetic model to synthesize kinetic data from static data, allowing for assessment of kinetic modeling accuracy. The phantom was designed with four main components: two left-ventricular (LV) myocardial sections and two LV blood-pool sections, sized for end-systole (ES) and end-diastole (ED). Each section of the phantom was imaged separately while acquiring list-mode data. These static, separate-compartment data were manipulated into synthetic dynamic data using a kinetic model representing the myocardium and blood-pool activity concentrations over time and then combined into a set of dynamic image frames and reconstructed. Regions of interest were drawn on the resulting images, and kinetic parameters were estimated. This process was performed for three tracer uptake values (K 1), three myocardial wall thicknesses, ten filter parameters, and 20 iterations for 25 noise ensembles. The degree of filtering and iteration number were optimized to minimize the root mean-squared error (RMSE) of K 1 values, with the largest number of iterations and minimal filtering yielding the lowest error. Using the optimized parameters, K 1 was determined with reasonable error (~3% RMSE) over all wall thicknesses and K 1 input values. This work demonstrates that accurate and precise measurements of K 1 are possible for the U-SPECT+ system used in this study, for several different uptake rates and LV dimensions. Additionally, it allows for future investigation utilizing other imaging systems, including PET studies with any radiotracer, as well as with additional phantom parts containing lesions.

Text mining applied to electronic cardiovascular procedure reports to identify patients with trileaflet aortic stenosis and coronary artery disease.

BACKGROUND: Interrogation of the electronic health record (EHR) using billing codes as a surrogate for diagnoses of interest has been widely used for clinical research. However, the accuracy of this methodology is variable, as it reflects billing codes rather than severity of disease, and depends on the disease and the accuracy of the coding practitioner. Systematic application of text mining to the EHR has had variable success for the detection of cardiovascular phenotypes. We hypothesize that the application of text mining algorithms to cardiovascular procedure reports may be a superior method to identify patients with cardiovascular conditions of interest. METHODS: We adapted the Oracle product Endeca, which utilizes text mining to identify terms of interest from a NoSQL-like database, for purposes of searching cardiovascular procedure reports and termed the tool "PennSeek". We imported 282,569 echocardiography reports representing 81,164 individuals and 27,205 cardiac catheterization reports representing 14,567 individuals from non-searchable databases into PennSeek. We then applied clinical criteria to these reports in PennSeek to identify patients with trileaflet aortic stenosis (TAS) and coronary artery disease (CAD). Accuracy of patient identification by text mining through PennSeek was compared with ICD-9 billing codes. RESULTS: Text mining identified 7115 patients with TAS and 9247 patients with CAD. ICD-9 codes identified 8272 patients with TAS and 6913 patients with CAD. 4346 patients with AS and 6024 patients with CAD were identified by both approaches. A randomly selected sample of 200-250 patients uniquely identified by text mining was compared with 200-250 patients uniquely identified by billing codes for both diseases. We demonstrate that text mining was superior, with a positive predictive value (PPV) of 0.95 compared to 0.53 by ICD-9 for TAS, and a PPV of 0.97 compared to 0.86 for CAD. CONCLUSION: These results highlight the superiority of text mining algorithms applied to electronic cardiovascular procedure reports in the identification of phenotypes of interest for cardiovascular research.

Relation of aortic valve calcium to chronic kidney disease (from the Chronic Renal Insufficiency Cohort Study).

Although subjects with chronic kidney disease (CKD) are at markedly increased risk for cardiovascular mortality, the relation between CKD and aortic valve calcification has not been fully elucidated. Also, few data are available on the relation of aortic valve calcification and earlier stages of CKD. We sought to assess the relation of aortic valve calcium (AVC) with estimated glomerular filtration rate (eGFR), traditional and novel cardiovascular risk factors, and markers of bone metabolism in the Chronic Renal Insufficiency Cohort (CRIC) Study. All patients who underwent aortic valve scanning in the CRIC study were included. The relation between AVC and eGFR, traditional and novel cardiovascular risk factors, and markers of calcium metabolism were analyzed using both unadjusted and adjusted regression models. A total of 1,964 CRIC participants underwent computed tomography for AVC quantification. Decreased renal function was independently associated with increased levels of AVC (eGFR 47.11, 44.17, and 39 ml/min/1.73 m2, respectively, p<0.001). This association persisted after adjusting for traditional, but not novel, AVC risk factors. Adjusted regression models identified several traditional and novel risk factors for AVC in patients with CKD. There was a difference in AVC risk factors between black and nonblack patients. In conclusion, our study shows that eGFR is associated in a dose-dependent manner with AVC in patients with CKD, and this association is independent of traditional cardiovascular risk factors.

Side-specific expression of activated leukocyte adhesion molecule (ALCAM; CD166) in pathosusceptible regions of swine aortic valve endothelium.

Aortic valve sclerosis (AVS), an early form of aortic valve disease, develops preferentially on the aortic side of valve leaflets, a predilection that is reflected in an heterogeneous side-specific gene expression profile. It has been ascertained that hypercholesterolemia is sufficient to initiate the endothelial expression of activated leukocyte adhesion molecule (ALCAM; CD166), restricted to the aortic side of the leaflet. Intercellular adhesion molecule-1 (ICAM-1) or vascular cell adhesion molecule-1 (VCAM-1)--both of which are more typically associated with early arterial inflammation--are not differentially expressed. ALCAM up-regulation by hypercholesterolemia suggests a side-specific spatial role in the recruitment of leukocytes to AVS sites.

Endothelial heterogeneity associated with regional athero-susceptibility and adaptation to disturbed blood flow in vivo.

Endothelial phenotypic heterogeneity plays an important role in the susceptibility of the cardiovascular system to disease. Arteries and heart valves are susceptible to chronic inflammatory disease in regions of blood flow disturbance that implicates hemodynamic forces and transport characteristics as prominent influences on endothelial phenotype. By combining in vivo high-throughput genomics (discovery science) and in vitro mechanistic approaches (reductionist science), we present endothelial patho-susceptibility as an imbalance of multiple interrelated pathways that sensitize the cells to pathological change. The recently identified association of endoplasmic reticulum stress with endothelium in regions of flow disturbance is outlined as an important example of susceptible phenotype linked to proinflammatory and oxidative stress pathways.

Hypercholesterolemia induces side-specific phenotypic changes and peroxisome proliferator-activated receptor-gamma pathway activation in swine aortic valve endothelium.

UNLABELLED: Background- The endothelium of healthy aortic valves expresses different phenotypes on the aortic and ventricular sides. On the aortic side, which is susceptible to aortic valve sclerosis, there is a balanced coexpression of both propathological and protective pathways. Side-specific global gene expression can address endothelial phenotype balance in early aortic valve sclerosis. METHODS AND RESULTS: Adult male swine were fed a hypercholesterolemic or an isocaloric normal diet for 2-week and 6-month periods. Hypercholesterolemia induced localized lipid insudation confined to the aortic side of the leaflet. Transcript profiling of valve endothelial populations showed that the susceptible aortic side was more sensitive to 2-week hypercholesterolemia than the ventricular side (1,325 vs 87 genes were differentially expressed). However, greater sensitivity was not evidence of a dysfunctional phenotype. Instead, pathway analyses identified differential expression of caspase 3-, peroxisome proliferator-activated receptor gamma-, TNF-alpha-, and nuclear factor-kappaB-related pathways that were consistent with a protective endothelial phenotype. This was confirmed at the protein level at 2 weeks and persisted at 6 months. CONCLUSIONS: In a large animal model at high spatial resolution, endothelium on the pathosusceptible side of the aortic valve leaflet is responsive to hypercholesterolemia. Transcript profiles indicative of a protective phenotype were induced and persisted on the side prone to aortic valve sclerosis.