Jeremy L. Warner

Using Topic Modeling via Non-negative Matrix Factorization to Identify Relationships between Genetic Variants and Disease Phenotypes: A Case Study of Lipoprotein(a) (LPA)

Genome-wide and phenome-wide association studies are commonly used to identify important relationships between genetic variants and phenotypes. Most of these studies have treated diseases as independent variables and suffered from heavy multiple adjustment burdens due to the large number of genetic variants and disease phenotypes. In this study, we propose using topic modeling via non-negative matrix factorization (NMF) for identifying associations between disease phenotypes and genetic variants. Topic modeling is an unsupervised machine learning approach that can be used to learn the semantic patterns from electronic health record data. We chose rs10455872 in LPA as the predictor since it has been shown to be associated with increased risk of hyperlipidemia and cardiovascular diseases (CVD). Using data of 12,759 individuals from the biobank at Vanderbilt University Medical Center, we trained a topic model using NMF from 1,853 distinct phecodes extracted from the cohort’s electronic health records and generated six topics. We quantified their associations with rs10455872 in LPA. Topics indicating CVD had positive correlations with rs10455872 (P < 0.001), replicating a previous finding. We also identified a negative correlation between LPA and a topic representing lung cancer (P < 0.001). Our results demonstrate the applicability of topic modeling in exploring the relationship between the genome and clinical diseases. Author summary Identifying the clinical associations of genetic variants remains crucial in understanding how the human genome modulates disease risk. Traditional phenome-wide association studies consider each disease phenotype as an independent variable, however, diseases often present as complex clusters of comorbid conditions. In this study, we propose using topic modeling to model electronic health record data as a mixture of topics (e.g., disease clusters or relevant comorbidities) and testing associations between topics and genetic variants. Our results demonstrated the feasibility of using topic modeling to replicate and discover novel associations between the human genome and clinical diseases.

Rare Variants in the Gene ALPL That Cause Hypophosphatasia Are Strongly Associated With Ovarian and Uterine Disorders

ContextnMutations in alkaline phosphatase (AlkP), liver/bone/kidney (ALPL), which encodes tissue-nonspecific isozyme AlkP, cause hypophosphatasia (HPP). HPP is suspected by a low-serum AlkP. We hypothesized that some patients with bone or dental disease have undiagnosed HPP, caused by ALPL variants.nnnObjectivenOur objective was to discover the prevalence of these gene variants in the Vanderbilt University DNA Biobank (BioVU) and to assess phenotypic associations.nnnDesignnWe identified subjects in BioVU, a repository of DNA, that had at least one of three known, rare HPP disease-causing variants in ALPL: rs199669988, rs121918007, and/or rs121918002. To evaluate for phenotypic associations, we conducted a sequential phenome-wide association study of ALPL variants and then performed a de-identified manual record review to refine the phenotype.nnnResultsnOut of 25,822 genotyped individuals, we identified 52 women and 53 men with HPP disease-causing variants in ALPL, 7/1000. None had a clinical diagnosis of HPP. For patients with ALPL variants, the average serum AlkP levels were in the lower range of normal or lower. Forty percent of men and 62% of women had documented bone and/or dental disease, compatible with the diagnosis of HPP. Forty percent of the female patients had ovarian pathology or other gynecological abnormalities compared with 15% seen in controls.nnnConclusionsnVariants in the ALPL gene cause bone and dental disease in patients with and without the standard biomarker, low plasma AlkP. ALPL gene variants are more prevalent than currently reported and underdiagnosed. Gynecologic disease appears to be associated with HPP-causing variants in ALPL.

SMART Cancer Navigator: A Framework for Implementing ASCO Workshop Recommendations to Enable Precision Cancer Medicine

PurposenData standards and interoperability are critical for improving care for patients with cancer. Recent efforts by ASCO include the Data Standards and Interoperability Summit in 2016, which led to the Omics and Precision Oncology and Advancing Interoperability workshops. To facilitate improved patient care, several recommendations for data sharing and standardization were made to the community.nnnMethodsnTo address these recommendations, we developed SMART Cancer Navigator, a Web application that uses application programming interfaces to gather clinical and genomic data from 11 public knowledge bases ranging from basic to clinical content coverage; three (CIViC, ClinVar, and OncoKB) explicitly linked genomic variants to clinical factors such as prognosis and treatment selection. We illustrated the utility of this application by selecting one of the monthly case studies presented by the ASCO University Molecular Oncology Tumor Board: Ovarian Cancer (BRCA Mutation). We also performed analyses on information from the three clinico-genomic knowledge bases to corroborate previous work and illustrate the state of data sharing among publicly available resources.nnnResultsnSMART Cancer Navigator aggregates and contextualizes data from 11 different knowledge bases and stores user queries in a lightweight Web application that can link into Fast Healthcare Interoperability Resources-enabled electronic health records. Potentially relevant clinical trials and/or approved treatments were identified for three mutations found in a hypothetical patient with advanced ovarian cancer. A comparison of the three clinico-genomic knowledge bases indicated substantial differences in coverage at the gene and variant levels.nnnConclusionnSMART Cancer Navigator has immediate relevance to practicing oncologists and others. Additional knowledge bases can be added without undue effort. As a first step toward utility, we generalized and disseminated the resulting implementation (https://smart-cancer-navigator.github.io) and data sets.

Computerized Approach to Creating a Systematic Ontology of Hematology/Oncology Regimens

PurposenThe systemic treatment of cancer is primarily through the administration of complex chemotherapy protocols. To date, this knowledge has not been systematized, because of the lack of a consistent nomenclature and the variation in which regimens are documented. For example, recording of treatment events in electronic health record notes is often through shorthand and acronyms, limiting secondary use. A standardized hierarchic ontology of cancer treatments, mapped to standard nomenclatures, would be valuable to a variety of end users.nnnMethodsnWe leveraged the knowledge contained in a large wiki of hematology/oncology drugs and treatment regimens, HemOnc.org. Through algorithmic parsing, we created a hierarchic ontology of treatment concepts in the World Wide Web Consortium Web Ontology Language. We also mapped drug names to RxNorm codes and created optional filters to restrict the ontology by disease and/or drug class.nnnResultsnAs of December 2017, the main ontology includes 30,526 axioms (eg, doxorubicin is an anthracycline), 1,196 classes (eg, regimens used in the neoadjuvant treatment of human epidermal growth factor receptor 2-positive breast cancer, nitrogen mustards), and 1,728 individual entities. More than 13,000 of the axioms are annotations including RxNorm codes, drug synonyms, literature references, and direct links to published articles.nnnConclusionnThis approach represents, to our knowledge, the largest effort to date to systematically categorize and relate hematology/oncology drugs and regimens. The ontology can be used to reason individual components from regimens mentioned in electronic health records (eg, R-CHOP maps to rituximab, cyclophosphamide, doxorubicin, vincristine, and prednisone) and also to probabilistically reconstruct regimens from individual drug components. These capabilities may be particularly valuable in the implementation of rapid-learning health systems on the basis of real-world evidence. The derived Web Ontology Language ontology is freely available for noncommercial use through the Creative Commons 4.0 Attribution-NonCommercial-ShareAlike license.

Overcoming the Straw Man Effect in Oncology: Visualization and Ranking of Chemotherapy Regimens Using an Information Theoretic Approach

Purpose Despite the plethora of randomized controlled trial (RCT) data, most cancer treatment recommendations are formulated by experts. Alternatively, network meta-analysis (NMA) is one method of analyzing multiple indirect treatment comparisons. However, NMA does not account for mixed end points or temporality. Previously, we described a prototype information theoretical approach for the construction of ranked chemotherapy treatment regimen networks. Here, we propose modifications to overcome an apparent straw man effect, where the most studied regimens were the most negatively valued. Methods RCTs from two scenarios—upfront treatment of chronic myelogenous leukemia and relapsed/refractory multiple myeloma—were assembled into ranked networks using an automated algorithm based on effect sizes, statistical significance, surrogacy of end points, and time since RCT publication. Vertex and edge color, transparency, and size were used to visually analyze the network. This analysis led to the additional incorporation of value propagation. Results A total of 18 regimens with 42 connections (chronic myelogenous leukemia) and 28 regimens with 25 connections (relapsed/refractory multiple myeloma) were analyzed. An initial negative correlation between vertex value and size was ameliorated after value propagation, although not eliminated. Updated rankings were in close agreement with published guidelines and NMAs. Conclusion Straw man effects can distort the comparative efficacy of newer regimens at the expense of older regimens, which are often cheaper or less toxic. Using an automated method, we ameliorated this effect and produced rankings consistent with common practice and published guidelines in two distinct cancer settings. These findings are likely to be generalizable and suggest a new means of ranking efficacy in cancer trials.

Abstract 1293: ABO blood type and cancer risk: preliminary findings from a phenome analysis

Introduction: ABO blood type has long been implicated in disease susceptibility, including cancer. However, evidence for associations with many malignancies is mixed. We applied a novel phenome approach to test to cancer codes from electronic medical records (EMR) in relation to ABO blood type in a large predominantly Caucasian study population. Approach: Among adults aged 18-100, cancer case and control status were assigned using 58 general neoplasm related phenome codes to de-identified EMR at the Vanderbilt University Medical Center. Blood type from serologic assays was ascertained from EMR-linked laboratory reports. Associations between blood type and cancer phenomes were quantified with Odds Ratios (OR) and corresponding 95% Confidence Intervals (CI) from logistic regression in models adjusted for sex and stratified by race/ethnicity. Only analyses with at least 100 cases per strata were conducted. Results: Among 221,015 Non-Hispanic Caucasians, 37,841 Blacks, 7,714 Hispanic Caucasians, and 3,616 Asian subjects with ABO blood type available in linked EMR, we evaluated 56, 37, 4, and 3 general cancer phenome codes, respectively. After employing Bonferroni corrections, ABO blood type was significantly associated with cancers of the pancreas, ovary, cervix, skin, and hematopoietic system. Caucasians with blood type O were less likely to have ovarian cancer (OR: 0.82, 95% CI 0.73-0.91) and pancreatic cancer (OR: 0.83, 95% CI: 0.74-0.92), and more likely to have squamous cell or other skin cancer (OR: 1.08, 95% CI: 1.04-1.13) and myeloid leukemia (OR: 1.15, 95% CI: 1.06-1.25) than those with other blood types (A, B, or AB). Hispanic Caucasians with blood type O were less likely to have cervical cancer (OR: 0.56, 95% CI: 0.38-0.82) than those with other blood types. No associations surpassed correction for multiple comparisons among Blacks or Asians. Conclusions: Our phenome approach confirmed known associations between blood type and risk of pancreatic and ovarian cancer, and adds to accumulating evidence supporting associations with skin cancer and leukemia. Our novel cervical cancer association among Hispanic Caucasians and other nominally significant findings, especially in understudied non-Caucasians, should be further evaluated in large and diverse populations. In addition, research to determine how ABO blood type may influence cancer development and progression, and if such associations can be exploited for risk prediction or cancer prevention is warranted. Citation Format: Alicia Beeghly-Fadiel, Ayush Giri, Lisa Bastarache, Jill Pulley, Jeremy Warner, Josh Denny. ABO blood type and cancer risk: preliminary findings from a phenome analysis [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2017; 2017 Apr 1-5; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2017;77(13 Suppl):Abstract nr 1293. doi:10.1158/1538-7445.AM2017-1293

Abstract P6-07-02: Targeted next generation sequencing of advanced breast cancers identifies potentially actionable alterations and variants of standard biomarkers in the majority of patients

Background: Molecular tumor profiling is increasingly important in the management of oncologic patients. Targeted next-generation sequencing (T-NGS), using formalin fixed paraffin embedded (FFPE) clinical samples, allows for molecular characterization of genes with known or potential therapeutic and prognostic importance in cancer. Actionable alterations include those with on- or off-label therapeutic implications, those that might be biomarkers of response to clinical trial agents, or those which are non-indicators for response. Design: We correlated information on patients with metastatic or refractory locoregional recurrence of breast cancer (BC) with potentially actionable genetic alterations detected by a commercially available T-NGS assay, which sequences the coding regions of 315 genes and introns of 28 genes involved in rearrangements selected for their demonstrated role in solid malignancy. We developed an informatics pipeline to capture test results in real-time and store them for subsequent research analysis. Results were analyzed by clinical subtype (estrogen receptor positive [ER+]; HER-2 amplified [HER2+]; triple-negative [TN]) for actionable alterations, most frequently altered genes/pathways and variants of standard biomarkers not detected by routine studies. Results: Between 11/2013 and 4/2015, 141 FFPE samples from 139 patients were tested by T-NGS. At least one potentially actionable genetic alteration was identified in 98% of patients (median 5.5 alterations/tumor [range 0-18]). 64% had alterations predicting sensitivity to approved BC therapies and 10% to approved therapies in other tumor types. An additional 1231 variants of uncertain significance (VUS) (median 9 per tumor [range 0-28]) were identified. The most frequently altered genes were TP53 (64%), PIK3CA (37%) and MYC (24%). Genes involved in cell cycle, DNA damage and PIK3CA/mTOR pathways were highly altered among all receptor subtypes. The RAS/MAPK pathway was more commonly altered in ER+ (28%) vs. HER2+ (13%) and TN (17%). CCND1 amplifications were found in 16% (57% ER+, 30% HER2+, 13% TN) and FGFR1 amplifications in 13% (61% ER+, 22% HER2, 17% TN). Co-amplification of 8p and 11q (including FGFR1/ZNF703 and CCND1/FGF3/FGF4/FGF19) was found in 28% of patients (ER+ 21%, HER2+ 25%, TN 7.5%). The combination of PIK3CA mutation and MAP3K1/MAP2K4 alteration occurred in 12% of patients (82% ER+, 18% TN). 4 ESR1 mutations and 2 amplifications (all in ER+) as well as 4 HER2 mutations (1 ER+ and 3 TN) were also identified. The number of patients receiving genotype-directed treatments informed by T-NGS results and patient outcome after genotype-directed treatment will be subsequently presented. Conclusion: Mutation profiling using T-NGS identified potentially actionable alterations in a majority of advanced BC patients, providing novel yet rational therapeutic options and facilitating clinical trial enrollment. T-NGS results will be used to guide therapy in increasing numbers of BC patients. Citation Format: Estrada MV, Warner J, Rioth M, Balko JM, Rexer B, Sanders ME. Targeted next generation sequencing of advanced breast cancers identifies potentially actionable alterations and variants of standard biomarkers in the majority of patients. [abstract]. In: Proceedings of the Thirty-Eighth Annual CTRC-AACR San Antonio Breast Cancer Symposium: 2015 Dec 8-12; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2016;76(4 Suppl):Abstract nr P6-07-02.