Diane T. Smelser

The Geisinger MyCode community health initiative: an electronic health record–linked biobank for precision medicine research

Purpose:Geisinger Health System (GHS) provides an ideal platform for Precision Medicine. Key elements are the integrated health system, stable patient population, and electronic health record (EHR) infrastructure. In 2007, Geisinger launched MyCode, a system-wide biobanking program to link samples and EHR data for broad research use.Methods:Patient-centered input into MyCode was obtained using participant focus groups. Participation in MyCode is based on opt-in informed consent and allows recontact, which facilitates collection of data not in the EHR and, since 2013, the return of clinically actionable results to participants. MyCode leverages Geisinger’s technology and clinical infrastructure for participant tracking and sample collection.Results:MyCode has a consent rate of >85%, with more than 90,000 participants currently and with ongoing enrollment of ~4,000 per month. MyCode samples have been used to generate molecular data, including high-density genotype and exome sequence data. Genotype and EHR-derived phenotype data replicate previously reported genetic associations.Conclusion:The MyCode project has created resources that enable a new model for translational research that is faster, more flexible, and more cost-effective than traditional clinical research approaches. The new model is scalable and will increase in value as these resources grow and are adopted across multiple research platforms.Genet Med 18 9, 906–913.

Genetic-based prediction of disease traits: Prediction is very difficult, especially about the future

Translation of results from genetic findings to inform medical practice is a highly anticipated goal of human genetics. The aim of this paper is to review and discuss the role of genetics in medically-relevant prediction. Germline genetics presages disease onset and therefore can contribute prognostic signals that augment laboratory tests and clinical features. As such, the impact of genetic-based predictive models on clinical decisions and therapy choice could be profound. However, given that (i) medical traits result from a complex interplay between genetic and environmental factors, (ii) the underlying genetic architectures for susceptibility to common diseases are not well-understood, and (iii) replicable susceptibility alleles, in combination, account for only a moderate amount of disease heritability, there are substantial challenges to constructing and implementing genetic risk prediction models with high utility. In spite of these challenges, concerted progress has continued in this area with an ongoing accumulation of studies that identify disease predisposing genotypes. Several statistical approaches with the aim of predicting disease have been published. Here we summarize the current state of disease susceptibility mapping and pharmacogenetics efforts for risk prediction, describe methods used to construct and evaluate genetic-based predictive models, and discuss applications.

Family Members of Patients with Abdominal Aortic Aneurysms are at Increased Risk for Aneurysms: Analysis of 618 Probands and their Families from the Liege AAA Family Study.

BACKGROUND The objectives were to answer the following questions with the help of a well-characterized population in Liège, Belgium: 1) what percentage of patients with abdominal aortic aneurysm (AAA) have a positive family history for AAA? 2) what is the prevalence of AAAs among relatives of patients with AAA? and 3) do familial and sporadic AAA cases differ in clinical characteristics? METHODS Patients with unrelated AAA diagnosed at the Cardiovascular Surgery Department, University Hospital of Liège, Belgium, between 1999 and 2012 were invited to the study. A detailed family history was obtained in interviews and recorded using Progeny software. We divided the 618 patients into 2 study groups: group I, 296 patients with AAA (268; 91% men) were followed up with computerized tomography combined with positron emission tomography; and group II, 322 patients with AAA (295; 92% men) whose families were invited to ultrasonographic screening. RESULTS In the initial interview, 62 (10%) of the 618 patients with AAA reported a positive family history for AAA. Ultrasonographic screening identified 24 new AAAs among 186 relatives (≥50 years) of 144 families yielding a prevalence of 13%. The highest prevalence (25%) was found among brothers. By combining the number of AAAs found by ultrasonographic screening with those diagnosed previously the observed lifetime prevalence of AAA was estimated to be 32% in brothers. The familial AAA cases were more likely to have a ruptured AAA than the sporadic cases (8% vs. 2.4%; P < 0.0001). CONCLUSIONS The findings confirm previously found high prevalence of AAA among brothers, support genetic contribution to AAA pathogenesis, and provide rationale for targeted screening of relatives of patients with AAA.

Population risk factor estimates for abdominal aortic aneurysm from electronic medical records: a case control study.

BackgroundUsing abdominal aortic aneurysm (AAA) as a model, this case–control study used electronic medical record (EMR) data to assess known risk factors and identify new associations.MethodsThe study population consisted of cases with AAA (n =888) and controls (n =10,523) from the Geisinger Health System EMR in Central and Northeastern Pennsylvania. We extracted all clinical and diagnostic data for these patients from January 2004 to December 2009 from the EMR. From this sample set, bootstrap replication procedures were used to randomly generate 2,500 iterations of data sets, each with 500 cases and 2000 controls. Estimates of risk factor effect sizes were obtained by stepwise logistic regression followed by bootstrap aggregation. Variables were ranked using the number of inclusions in iterations and P values.ResultsThe benign neoplasm diagnosis was negatively associated with AAA, a novel finding. Similarly, type 2 diabetes, diastolic blood pressure, weight and myelogenous neoplasms were negatively associated with AAA. Peripheral artery disease, smoking, age, coronary stenosis, systolic blood pressure, age, height, male sex, pulmonary disease and hypertension were associated with an increased risk for AAA.ConclusionsThis study utilized EMR data, retrospectively, for risk factor assessment of a complex disease. Known risk factors for AAA were replicated in magnitude and direction. A novel negative association of benign neoplasms was identified. EMRs allow researchers to rapidly and inexpensively use clinical data to expand cohort size and derive better risk estimates for AAA as well as other complex diseases.

Long-Term Quality of Life of Abdominal Aortic Aneurysm Patients Under Surveillance or After Operative Treatment

BACKGROUND The objective of this study was to determine the long-term quality of life (QOL) in patients with an abdominal aortic aneurysm (AAA) undergoing surveillance or after operative treatment. METHODS 249 patients with AAAs completed the WHO Quality of Life-BREF (WHOQOL-BREF) test and Short Form (36) Health Survey (SF-36) survey: 78 patients with small AAAs under surveillance, 26 after ruptured AAAs (rAAAs), 47 after endovascular aneurysm repair (EVAR), and 98 after elective open repair. The results were compared with WHOQOL-BREF and SF-36 standard values from a matched German population using the Students 2-tailed t-test. RESULTS Long-term results of the WHOQOL-BREF test showed that patients undergoing AAA surveillance had a significantly lower physical QOL (P = 0.04). Patients after EVAR or open repair rated their environmental QOL significantly higher than the age- and sex-matched general population (open repair: P = 0.006; EVAR: P < 0.001). Patients with rAAAs had the same QOL as the matched German population. Long-term results of the QOL SF-36 showed that patients undergoing AAA surveillance rated their QOL significantly lower in the subgroup of role-physical (P = 0.02) and role-emotional (P = 0.003). Patients with rAAAs rated lower scores for role-physical (P = 0.02) and had more bodily pain (P = 0.02). Patients who underwent elective open repair had the same high QOL as the matched German population, whereas patients who underwent EVAR reported significant improvement in vitality (P = 0.002) and mental health (P = 0.03) compared with the matched German population. CONCLUSIONS Based on measurements from 2 independent QOL tests, the well-established operative treatment of AAAs provided patients with a QOL comparable to that of a matched German population. The electively treated AAA groups rated environmental QOL factors significantly higher than the control group. The impaired physical and emotional QOL of the AAA group under surveillance suggests that more intense patient education could be beneficial.

ePhenotyping for Abdominal Aortic Aneurysm in the Electronic Medical Records and Genomics (eMERGE) Network: Algorithm Development and Konstanz Information Miner Workflow.

Background and objective We designed an algorithm to identify abdominal aortic aneurysm cases and controls from electronic health records to be shared and executed within the “electronic Medical Records and Genomics” (eMERGE) Network. Materials and methods Structured Query Language, was used to script the algorithm utilizing “Current Procedural Terminology” and “International Classification of Diseases” codes, with demographic and encounter data to classify individuals as case, control, or excluded. The algorithm was validated using blinded manual chart review at three eMERGE Network sites and one non-eMERGE Network site. Validation comprised evaluation of an equal number of predicted cases and controls selected at random from the algorithm predictions. After validation at the three eMERGE Network sites, the remaining eMERGE Network sites performed verification only. Finally, the algorithm was implemented as a workflow in the Konstanz Information Miner, which represented the logic graphically while retaining intermediate data for inspection at each node. The algorithm was configured to be independent of specific access to data and was exportable (without data) to other sites. Results The algorithm demonstrated positive predictive values (PPV) of 92.8% (CI: 86.8-96.7) and 100% (CI: 97.0-100) for cases and controls, respectively. It performed well also outside the eMERGE Network. Implementation of the transportable executable algorithm as a Konstanz Information Miner workflow required much less effort than implementation from pseudo code, and ensured that the logic was as intended. Discussion and conclusion This ePhenotyping algorithm identifies abdominal aortic aneurysm cases and controls from the electronic health record with high case and control PPV necessary for research purposes, can be disseminated easily, and applied to high-throughput genetic and other studies.

Adiposity, inflammation, genetic variants and risk of post-menopausal breast cancer findings from a prospective-specimen-collection, retrospective- blinded-evaluation (PRoBE) design approach

Chronic internal inflammation secondary to adiposity is a risk factor for sporadic breast cancer and Post-Menopausal Breast Cancer (PMBC) is largely defined as such. Adiposity is one of the clinical criteria for the diagnosis of Metabolic Syndrome (MetS) and is a risk factor for PMBC. We examined SNPs of eight genes implicated in adiposity, inflammation and cell proliferation in a Prospective-specimen-collection, Retrospective-Blinded-Evaluation (PRoBE) design approach. A total of 180 cases and 732 age-matched controls were identified from the MyCode prospective biobank database and then linked to the Clinical Decision Information System, an enterprise-wide data warehouse, to retrieve clinico-demographic data. Samples were analyzed in a core laboratory where the personnel were masked to their status. Results from multivariate logistic regression yielded one SNP (rs2922126) in the GHSR as protective against PMBC among homozygotes for the minor allele (A/A) (OR = 0.4, 95% CI 0.18-.89, P-value = .02); homozygosity for the minor allele (C/C) of the SNP (rs889312) of the gene MAP3K1 was associated with the risk of PMBC (OR = 2.41, 95% CI 1.25-4.63 P-value = .008). Advanced age was protective against PMBC (OR = 0.98, 95% CI 0.95-0.99, P-value = .02). Family history of breast cancer (OR = 2.22, 95% CI 1.14-4.43. P = .02), HRT (OR = 3.35; 95% CI 2.15-5.21, P < .001), and MetS (OR = 14.83, 95% CI 5.63-39.08, P < .001) and interaction between HRT and MetS (OR = 39.38, 95% CI 15.71-98.70, P < .001) were associated with the risk of PMBC. We did not detected significant interactions between SNPs or between the SNPs and the clinico-demographic risk factors. Our study further confirms that MetS increases the risk of PMBC and argues in favor of reducing exposure to HRT. Our findings are another confirmation that low penetrance genes involved in the inflammatory pathway, i.e. MAP3KI gene, may have a plausible causative role in PMBC. Given the fact that genetic constitutionality of individuals cannot be changed, efforts should be focused on life style modification.

Genetic risk models: Influence of model size on risk estimates and precision

Disease risk estimation plays an important role in disease prevention. Many studies have found that the ability to predict risk improves as the number of risk single‐nucleotide polymorphisms (SNPs) in the risk model increases. However, the width of the confidence interval of the risk estimate is often not considered in the evaluation of the risk model. Here, we explore how the risk and the confidence interval width change as more SNPs are added to the model in the order of decreasing effect size, using both simulated data and real data from studies of abdominal aortic aneurysms and age‐related macular degeneration. Our results show that confidence interval width is positively correlated with model size and the majority of the bigger models have wider confidence interval widths than smaller models. Once the model size is bigger than a certain level, the risk does not shift markedly, as 100% of the risk estimates of the one‐SNP‐bigger models lie inside the confidence interval of the one‐SNP‐smaller models. We also created a confidence interval‐augmented reclassification table. It shows that both more effective SNPs with larger odds ratios and less effective SNPs with smaller odds ratios contribute to the correct decision of whom to screen. The best screening strategy is selected and evaluated by the net benefit quantity and the reclassification rate. We suggest that individuals whose upper bound of their risk confidence interval is above the screening threshold, which corresponds to the population prevalence of the disease, should be screened.

PS2-8: ePhenotyping for Abdominal Aortic Aneurysm

Background/Aims A large volume of clinical data is captured in electronic medical records (EMRs), and feasibly extracting the data to define clinical phenotypes is valuable to health care research. We designed an algorithm to define abdominal aortic aneurysm (AAA) cases and controls. We implemented the algorithm using our institutional warehouse and propose using the HMORN Virtual Data Warehouse (VDW) to replicate our findings. Methods The cohort consisted of individuals enrolled in the Geisinger MyCode biobank or consented for research in other studies (such as the Vascular Department). The Structured Query Language (SQL) algorithm utilized CPT codes and ICD9 codes and vital signs data to define individuals as cases, controls or excludes. AAA cases were defined as having an AAA repair procedure, or at least one vascular clinic encounter with a ruptured AAA, or at least two vascular clinic encounters with an unruptured AAA. AAA controls were neither excludes nor cases, had an encounter within the past 5 years, and never had an ICD9 code 441.3, 441.4, or 441.9. Individuals were excluded based on certain medical conditions, age younger than 40 or older than 89, not having an encounter within 5 years, or having an ICD9 diagnosis of 441. Results We screened the records of 29,770 individuals, identifying 1,155 AAA cases and 17,523 controls. We excluded 337 individuals based on predisposing genetic conditions, 109 individuals without a visit within the past 5 years and 10,398 individuals based on age. To assure that we had true AAA cases, 248 individuals with ICD-9 codes of 441.x (which includes thoracic and unspecified site of aneurysm) were excluded. The algorithm was validated on a subset of individuals by manual chart review and demonstrated a Positive Predictive Value (PPV) of 94% and sensitivity of 100%. Conclusions We designed an ePhenotyping algorithm to identify AAA cases and controls from the EMR with high PPV and sensitivity necessary for research purposes. The VDW provides an excellent opportunity to broaden the study population characteristics and replicate the findings.

PS2-41: The FTO Gene, Its Variants and Post-menopausal Breast Cancer

Background Adiposity is a well-accepted risk factor for post-menopausal breast cancer (PMBC). SNPs of the FTO gene, located on chromosome 16q12.2, have been associated with adiposity by lowering the activity of the enzyme lipase. This gene also encodes for a protein involved in DNA damage repair pathway. The current epidemiologic paradigm associates the risk of PMBC with increased aromatization of androstenedione to estradial in peripheral adipose tissue and increased serum levels of insulin-like growth-factors. The increased systemic inflammation because of adiposity motivates the inquiries for other plausible biochemical mechanism. We are conducting a pilot nested case-controls study to investigate the association between 3 SNPs of FTO gene with the risk of PMBC, adjusted for the major risk factors. Presently data are available from 50 cases and 202 controls from the cohort of MyCode Biobanking Project at Geisinger Health System. Methods We genotyped the 3 SNPs (rs1861868, rs1477196 and rs9939609) of the gene. Generalized linear model was used to evaluate the association between these 3 SNPs. We also conducted analyses of hyplotypes for 2-SNPs and 3-SNPs association, respectively; odds ratios (OR) and 95% confidence intervals (CI) were calculated to estimate the association between haplotypes and PMBC, and between the genotypes of each SNPs (as an ordinal variable) and PMBC applying unconditional logistic regression modeling approach, while controlling for BMI, age, parity status (=3 or <3) and family history for breast cancer. Results Our analyses suggest that 2 of the SNPs (rs1861868 and rs1477196) most likely form a haplotype (P= 0.002) that is exclusive of the third SNP (rs9939609, P>0.3). None of these SNPs or hyplotypes was associated with BMI (p>0.4), diabetes and metabolic syndrome (P>0.15). Results from the multivariate logistic model suggested a trend toward a potential association (OR=1.68, CI=0.67–4.18, P=0.09) between the (rs1861868) SNP and PMBC. Conclusion Our pilot study attempts to evaluate the association between SNPs of the of FTO gene with PMBC, independent of BMI or other risk factors for PMBC. One SNP was found to be likely associated with PMBC, but the present small sample size eclipses the power of identifying a definite association.