Citations

Abstract Objective Health care generated data have become an important source for clinical and genomic research. Often, investigators create and iteratively refine phenotype algorithms to achieve high positive predictive values (PPVs) or sensitivity, thereby identifying valid cases and controls. These algorithms achieve the greatest utility when validated and shared by multiple health care systems. Materials and Methods We report the current status and impact of the Phenotype KnowledgeBase (PheKB, http://phekb.org ), an online environment supporting the workflow of building, sharing, and validating electronic phenotype algorithms. We analyze the most frequent components used in algorithms and their performance at authoring institutions and secondary implementation sites. Results As of June 2015, PheKB contained 30 finalized phenotype algorithms and 62 algorithms in development spanning a range of traits and diseases. Phenotypes have had over 3500 unique views in a 6-month period and have been reused by other institutions. International Classification of Disease codes were the most frequently used component, followed by medications and natural language processing. Among algorithms with published performance data, the median PPV was nearly identical when evaluated at the authoring institutions (n = 44; case 96.0%, control 100%) compared to implementation sites (n = 40; case 97.5%, control 100%). Discussion These results demonstrate that a broad range of algorithms to mine electronic health record data from different health systems can be developed with high PPV, and algorithms developed at one site are generally transportable to others. Conclusion By providing a central repository, PheKB enables improved development, transportability, and validity of algorithms for research-grade phenotypes using health care generated data.

The legacy of human-Neandertal interbreeding Non-African humans are estimated to have inherited on average 1.5 to 4% of their genomes from Neandertals. However, how this genetic legacy affects human traits is unknown. Simonti et al. combined genotyping data with electronic health records. Individual Neandertal alleles were correlated with clinically relevant phenotypes in individuals of European descent. These archaic genetic variants were associated with medical conditions affecting the skin, the blood, and the risk of depression. Science , this issue p. 737

Abstract Purpose: A prophylactic ureteral localization stent (PULSe) placed by urologists aids in intraoperative localization and detection of suspected ureteral injury during complex colorectal surgery (CRS) cases. We evaluated the incidence and management of urologic-induced complications secondary to PULSe placement during CRS cases at a single center. Materials and Methods: We performed a retrospective review of all patients who underwent cystoscopy and PULSe placement at the time of CRS over a 12-month period. Bilateral 5 French ×70-cm TigerTail ® (Bard Medical Division, Covington, GA) PULSe devices were placed without assistance of routine fluoroscopy. Results: Ninety-nine patients (mean age, 58.1 years; range, 17–88 years) underwent bilateral PULSe placement, with a male:female ratio of 44:55 and a mean body mass index of 26.8 (17.0–38.6) kg/m 2 . Mean pre- and postprocedural creatinine levels were 0.91 and 1.01 mg/dL, respectively. Twenty-two of 99 (22%) cases utilized a guidewire to aid in placement of PULSe. Four Clavien grade IIIb complications occurred: mucosal edema, reflex anuria, ureteral perforation, and ureteral obstruction secondary to significant clot burden. Three of the grade IIIb complications were managed endoscopically with double-J stent placement. The ureteral perforation case required percutaneous nephrostomy tube placement. Subgroup analysis of the four grade IIIb complications revealed a mean age of 62.3 years, body mass index of 26.98 kg/m 2 , and pre- and postprocedural creatinine levels of 0.95 and 4.83 mg/dL, respectively. Only one of the four grade IIIb complications utilized a guidewire prior to PULSe placement. Conclusions: The incidence of Clavien grade III urologic-induced complications during PULSe placement is approximately 2% (4/188). Mandatory adoption of fluoroscopy and guidewires may be required to minimize complications of PULSe placement.

AbstractBackground Electronic health records (EHRs) are increasingly used for clinical and translational research through the creation of phenotype algorithms. Currently, phenotype algorithms are most commonly represented as noncomputable descriptive documents and knowledge artifacts that detail the protocols for querying diagnoses, symptoms, procedures, medications, and/or text-driven medical concepts, and are primarily meant for human comprehension. We present desiderata for developing a computable phenotype representation model (PheRM).Methods A team of clinicians and informaticians reviewed common features for multisite phenotype algorithms published in PheKB.org and existing phenotype representation platforms. We also evaluated well-known diagnostic criteria and clinical decision-making guidelines to encompass a broader category of algorithms.Results We propose 10 desired characteristics for a flexible, computable PheRM: (1) structure clinical data into queryable forms; (2) recommend use of a common data model, but also support customization for the variability and availability of EHR data among sites; (3) support both human-readable and computable representations of phenotype algorithms; (4) implement set operations and relational algebra for modeling phenotype algorithms; (5) represent phenotype criteria with structured rules; (6) support defining temporal relations between events; (7) use standardized terminologies and ontologies, and facilitate reuse of value sets; (8) define representations for text searching and natural language processing; (9) provide interfaces for external software algorithms; and (10) maintain backward compatibility.Conclusion A computable PheRM is needed for true phenotype portability and reliability across different EHR products and healthcare systems. These desiderata are a guide to inform the establishment and evolution of EHR phenotype algorithm authoring platforms and languages.

Abstract Background: The use of anti-estrogen (AE) therapies (selective estrogen receptor modulators [SERMs], aromatase inhibitors [AIs]) in breast cancer (BC) patients was recently associated with risk of developing rheumatoid arthritis (RA) (J Rheum 2015;42:55-9). We attempted to replicate and extend these results using electronic health record (EHR) enhanced cancer registry data that assessed specific AE therapies and also accounted for potential confounding by age, smoking status, and chemotherapy use. Methods: Using cancer registry and EHR data, we identified a cohort of BC patients who were newly diagnosed and treated at Mayo Clinic Rochester from 1998-2011 and had no previous diagnosis of RA. Smoking status at diagnosis, BC treatments and RA at diagnosis and during follow-up were identified using validated EHR-based algorithms. Hazard ratios (HRs) and 95% confidence intervals (CI) from a multivariate Cox model were used to estimate the association of AE use with risk of RA, controlling for age, smoking status at BC diagnosis, and BC chemotherapy use. Results: The analytic cohort of 9,244 newly diagnosed BC patients treated and followed at Mayo Clinic had a median age at diagnosis of 59 years (range 18-97); 32% were smokers; and 29% were treated with chemotherapy. During a median follow-up of 49 months (range 1-191), 19 patients developed RA. Compared to BC patients not receiving any AE therapy, those receiving AI monotherapy, but not SERM monotherapy or both SERMs and AIs, were at significantly increased risk of developing RA (Table). AE Therapy# of RA Cases / # patientsHR* (95% CI)P-valueNo AE therapy6/4,0711.00 (reference)SERMs only5/2,5371.20 (0.36 - 3.95)0.77AIs only5/7126.37 (1.77 - 22.92)0.005SERMs then AIs3/1,8750.90AI then SERMs0/490.91 (0.21 - 3.86)*Adjusted for age, smoking status and use of chemotherapy Conclusions: Our findings indicate that the use of AIs for BC therapy, but not SERMs or SERMs followed by AIs, is associated with increased risk for development of RA, when controlling for smoking status and use of chemotherapy. While needing replication, these findings suggest a specific role for AIs in RA development and suggest pathways that could be targeted to prevent this potential treatment complication. Citation Format: Matthew K. Breitenstein, Ming-Fen Ho, Richard M. Weinshilboum, James R. Cerhan, Jyotishman Pathak, Tim Bongartz, Liewei Wang, James N. Ingle. Association of anti-estrogen therapy in breast cancer patients and subsequent risk of rheumatoid arthritis: An electronic health record study. [abstract]. In: Proceedings of the 106th Annual Meeting of the American Association for Cancer Research; 2015 Apr 18-22; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2015;75(15 Suppl):Abstract nr LB-191. doi:10.1158/1538-7445.AM2015-LB-191

Abstract Epithelial ovarian cancer (EOC) is the fifth leading cause of cancer death among women in the United States (5% of cancer deaths). The standard treatment for patients with advanced EOC is initial debulking surgery followed by carboplatin-paclitaxel combination chemotherapy. Unfortunately, even with modern chemotherapy, most patients relapse and die with the five-year overall survival around 45%. In addition, those patients that initially respond to taxane-platinum therapy eventually develop platinum-resistant tumors and relapse. Thus, finding novel therapeutics for treating EOC is essential. One approach that has been used widely in cancer drug discovery is Connectivity Mapping (CMAP) using gene expression data and drug phenotype data assessed on cancer cell lines (CCL) (i.e. “sensitive” or “resistant” based on dose-response drug data). However, recent publications have highlighted some issues with the use of these standard CCL sets that are often used in these drug screening studies. Therefore, instead of CMAP analysis based on gene expression signatures developed on drug response data collected on a small set of EOC cell lines, we took a novel CMAP approach based on information collected on EOC patients and clinical endpoints. That is, we determined the gene expression signature (e.g., set of genes) that were associated with time to recurrence in data collected on EOC patients within TCGA (n = 518) and a Mayo Clinic study (n = 474). The set of genes associated with time to recurrence (with and without adjustment for additional clinical covariates) was completed separately for TCGA and Mayo Clinic studies, restricting all analyses to EOC tumor of high-grade serous histology. Each of these 4 sets of genes, where genes with hazard ratios (HR) > 1 coded as “positively” associated and genes with HR < 1 coded as “negatively” associated with the phenotype, were inputted into CMAP software (Broad Institute) to determine a set of drugs for which our signature “matches” the “reference” signature. The drugs that overlapped between the CMAP analyses on the two studies were carried forward for validation studies involving drug screens on a set of 10 EOC cell lines. Of the 11 drugs carried forward, we a priori hypothesized that 5 of these drugs would kill EOC cells (mitoxantrone, podophyllotoxin, wortmannin, doxorubicin, and 17-AAG), for which all 5 showed an ability to kill EOC cells in vitro (cell lines were treated using serial dilutions of the drugs (20-0 μM) for 72 h followed by measuring cell viability using the CellTiter-Blue assay). Further research is on-going to understand the functional mechanism by which these candidate drugs might treat EOC and to apply this bioinformatics approach to other cancer drug screening studies. Note: This abstract was not presented at the meeting. Citation Format: Brooke L. Fridley, Rama Raghavan, Gottfried E. Konecny, Chen Wang, Ellen L. Goode, Harsh B. Pathak, Stephen Hyter. A bioinformatics approach to drug discovery: patient based connectivity mapping. [abstract]. In: Proceedings of the 106th Annual Meeting of the American Association for Cancer Research; 2015 Apr 18-22; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2015;75(15 Suppl):Abstract nr 4853. doi:10.1158/1538-7445.AM2015-4853

Abstract Objective To review and evaluate available software tools for electronic health record–driven phenotype authoring in order to identify gaps and needs for future development. Materials and Methods Candidate phenotype authoring tools were identified through (1) literature search in four publication databases (PubMed, Embase, Web of Science, and Scopus) and (2) a web search. A collection of tools was compiled and reviewed after the searches. A survey was designed and distributed to the developers of the reviewed tools to discover their functionalities and features. Results Twenty-four different phenotype authoring tools were identified and reviewed. Developers of 16 of these identified tools completed the evaluation survey (67% response rate). The surveyed tools showed commonalities but also varied in their capabilities in algorithm representation, logic functions, data support and software extensibility, search functions, user interface, and data outputs. Discussion Positive trends identified in the evaluation included: algorithms can be represented in both computable and human readable formats; and most tools offer a web interface for easy access. However, issues were also identified: many tools were lacking advanced logic functions for authoring complex algorithms; the ability to construct queries that leveraged un-structured data was not widely implemented; and many tools had limited support for plug-ins or external analytic software. Conclusions Existing phenotype authoring tools could enable clinical researchers to work with electronic health record data more efficiently, but gaps still exist in terms of the functionalities of such tools. The present work can serve as a reference point for the future development of similar tools.

Abstract Background: Literature addressing metformin exposure post diagnosis of breast cancer hints at a potential protective association on mortality outcomes, although findings have been inconclusive. In this research, we contribute to this body of evidence by using robust clinical indicators to focus on subgroups of breast cancer patients that might be most appropriate for potential metformin treatment. Methods: Robust breast cancer staging, treatment, and medication exposure attributes were compiled from Mayo Clinic Electronic Health Records and Breast Cancer Registry using a combination of structured queries and natural language processing of clinical notes. A cohort of female Breast Cancer Registry patients having type II diabetes mellitus (T2DM) at breast cancer diagnosis with a median follow-up period of 79.0 months after breast cancer diagnosis was identified (N=350). Univariate and multivariate cox proportional hazard regression analysis were used to model characteristics of BMI at diagnosis, age at diagnosis, metformin, insulin, thiazolidinedione, and sulfonylurea exposure after diagnosis, cancer grade, cancer stage, chemotherapy, hormone therapy, radiotherapy, lumpectomy, and mastectomy. Patients presenting with metastatic breast cancer at diagnosis, receiving immunotherapy for breast cancer treatment, and concurrent metformin and insulin exposure for T2DM treatments were excluded from the analysis cohort. Results: Univariate analysis independently identified metformin exposure (HR=0.243,p=0.0016), insulin exposure (HR=3.481,p=0.0210), age at breast cancer diagnosis (HR=1.075,p=0.0001), and targeted radiotherapy (HR=0.337,p=0.0082) as having a significant association with all-cause mortality. The multivariate model of metformin exposure (HR=0.274,p=0.0123), insulin exposure (HR=2.502,p=0.1160), age at breast cancer diagnosis (HR=1.043,p=0.0480), and targeted radiotherapy (HR=0.303,p=0.0044) on all-cause mortality demonstrated significant associations for all variables (Wald Chi-Square=25.2916,p<0.0001). A parallel multivariate model demonstrated metformin exposure (HR=0.101,p=0.0323), insulin exposure (HR=5.241,p=0.0265), and targeted radiotherapy (HR=0.292,p=0.0520) as having significant associations with breast cancer-specific mortality. Discussion: These findings suggest a protective association for metformin and a potential detrimental association for insulin therapy on all-cause and breast cancer-specific mortality amongst non-metastatic breast cancer patients with T2DM. The findings presented include robust clinical indicators for BMI, stage, and grade at diagnosis, post-diagnosis diabetes medication exposure, and endpoints of all cause and breast cancer specific mortality that have not been available in larger population studies. Our findings have potential pharmacogenetic implications for inhibition of localized breast cancer cell growth by targeting downstream components of the adenosine monophosphate-activated protein kinase pathway, the insulin/insulin-like growth factor-1 signaling pathway, and directly and indirectly via additional metabolic pathways with specific drug therapy combinations. Citation Format: Matthew K. Breitenstein, Leiwei Wang, Richard M. Weinshilboum, Jyotishman Pathak. Metformin exposure on survival in localized breast cancer patients with type II diabetes mellitus using electronic health records. [abstract]. In: Proceedings of the 105th Annual Meeting of the American Association for Cancer Research; 2014 Apr 5-9; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2014;74(19 Suppl):Abstract nr LB-278. doi:10.1158/1538-7445.AM2014-LB-278