Petterson, Tanya M.

Abstract If the observed dramatic increase in VTE incidence with advancing age is due to increased VTE risk factor exposure (i.e., secondary VTE), the incidence of idiopathic VTE should not vary with age or calendar year. Objective: To estimate the incidence of idiopathic and secondary VTE by age and by calendar year. Methods: Using the resources of the Rochester Epidemiology Project, we identified the inception cohort of Olmsted County, MN, residents with a first lifetime VTE during the 30-year period, 1966–1995 (n=2761). For each case, we reviewed the complete medical records in the community for 48 baseline clinical characteristics that are commonly-accepted risk factors for VTE. We categorized VTE cases as idiopathic (n=305) if no such characteristics were present; the remaining cases were categorized as secondary. Age- and sex-specific incidence rates were calculated using idiopathic or secondary VTE cases as the numerator, and age-, sex- and calendar year-specific estimates of the population of Olmsted County as the denominator. Results: The incidence of both idiopathic and secondary VTE increased exponentially with age for both men and women (p<0.001). Over the 30-year study period, the age-adjusted incidence of idiopathic VTE was essentially constant among men (from 19.0 to 17.1 per 100,000 men-years for 1966–70 and 1990–95, respectively), but decreased markedly among women (from 18.5 to 3.6 per 100,000 woman-years for 1966–70 and 1990–95, respectively; p=0.005 for the interaction). Conclusions: The dramatic increase in VTE incidence with age likely reflects the biology of aging, although as yet unidentified VTE risk factors cannot be excluded. The reason for the decreased incidence of idiopathic VTE over the last 30 years among women is unclear.

Abstract Abstract 1241 Background: Hospitalization (with or without surgery) is a major risk factor for incident venous thromboembolism (VTE); however, the contribution of interim hospitalization to risk of recurrent VTE is unknown. Objective: To estimate risk of recurrent VTE related to interim hospitalization by conducting a population-based longitudinal review of provider-linked detailed medical records. Methods: We performed a nested case-cohort study. The cohort consisted of all Olmsted County residents with incident VTE 1988–2000 and ≥1 day follow-up. Cases were cohort members with recurrent VTE. Subjects were followed for all interim hospitalizations and warfarin use from incident VTE until earliest of emigration, death, recurrent VTE, or 12/31/2005. Data were analyzed using Cox proportional hazards and time dependent covariates to test for the effects of interim hospitalization and prophylaxis on VTE recurrence, adjusting for gender and age at incident VTE. Analyses were limited to subjects who survived free of death and recurrent VTE for ≥ 6 months. Results: Of 1262 incident VTE events (cohort), there were 309 VTE recurrences (cases). We randomly sampled 272 subjects from the cohort and 163 cases. Of the random samples, 210 incident events and 83 cases survived ≥ 6 months free of death and recurrent VTE and form our analysis population. The rate of secondary (interim) prophylaxis was approximately 50% for both incident events and cases, and was not predictive of recurrence (p=0.73). Male gender and interim hospitalization were associated with increased VTE recurrence even after adjusting for age at incident VTE and use of secondary warfarin prophylaxis. The hazard of recurrent VTE was nearly 10-fold higher for subjects with interim hospitalization versus those with none (HR: 9.6; 95% CI: 6.6, 13.8); men had a 1.5-fold increased recurrence rate compared with women (HR: 1.5; 95% CI: 1.1, 2.1). Conclusions: Our results, for the first time, show the importance of interim hospitalization as a predictor of VTE recurrence. Disclosures: No relevant conflicts of interest to declare.

Abstract Active cancer is an independent VTE risk factor (overall 6- to 9-fold increased risk) and accounts for almost 20% of all VTE in the community, but which cancer patients are at risk for VTE is largely unknown. Reportedly, VTE risk varies by tumor site, and cancer of the ovary, pancreas, colon, stomach, lung, prostate, and kidney convey particularly high VTE risk. Objective: To estimate VTE risk by tumor site. Methods: We enumerated observed cancers by tumor site for Olmsted County, MN active cancer patients with incident VTE over the seven-year period, 1991–1997 (n=152). We used 1991–1997 State Surveillance, Epidemiology, and End Results (SEER) data for Iowa to estimate the expected age-specific prevalence of cancer by tumor site in Olmsted County. VTE risk ratios (RR) for each tumor site were estimated by dividing the observed number of cancers by the expected number (calculated as the product of the SEER prevalence and the number of incident VTE cases in the age stratum). Results: For our population of 1991–1997 VTE cases, all tumor sites had RR > 5.0 (range 5.2 to 37.3, all p-values<0.05). Compared to published overall VTE odds ratios of 6–9 for active cancer compared to no cancer, the RR for some tumor sites were particularly increased. A Chi-squared test of heterogeneity of the RR across sites was highly significant (p-value<0.001). Three rare cancer sites - pancreatic cancer, lymphoma, and brain cancer - had unusually high RR (all RR>25). The high number of VTE cases with lymphoma was not due to catheter-related arm vein thrombosis. Liver, leukemia, other gastrointestinal (esophagus, small intestine, gallbladder, other biliary) and other gynecologic (primarily cervical) cancers had over twice the baseline risk (i.e., RR>17.0). On the other hand, the RR for many common cancers (breast, colorectal, ovary, lung, prostate) were essentially the same as the overall baseline risk (all had 9.5<RR<12.0). Conclusions: In contrast to previous reports, pancreas, lymphoma, brain, liver, leukemia, other gastrointestinal, and other gynecologic cancers have the highest VTE risk. Prior estimates of VTE risk by tumor site may have been biased by studies of prevalent cancers among patients hospitalized in tertiary care centers.

Abstract While atherosclerosis has been suggested as a risk factor for VTE, previously identified mechanisms for atherosclerosis and VTE differ dramatically. Objectives: To compare (a) VTE-free survival and (b) observed vs. expected number of VTE events, among adult patients with and without aortic atherosclerosis. Design/Setting: Population-based prospective cohort study in Olmsted County, MN. Patients: Age- and sex-stratified random sample of the adult (≥45 years old) Olmsted County population identified over the three-year period, 1993–1995 (n=573), performed as part of the Stroke Prevention: Assessment of Risk in the Community (SPARC) study. Measurements: The SPARC study included a baseline clinical evaluation (interviews, review of medical records), blood pressure measurements, blood sampling, and transesophageal echocardiography (TEE). The Risk Factors for VTE in the Community study identified all incident VTE among Olmsted County residents over the 32-year period, 1966–1997, from which expected VTE events by age and sex were calculated. Results: Over 2830 person-years of follow-up, 12 SPARC patients developed VTE (7 idiopathic). The 4-year VTE-free survival was 99% and 98% for those with and those without any TEE-detected aortic atherosclerosis. Adjusting for age in a Cox proportional hazards model, lack of aortic atherosclerosis was significantly associated with subsequent VTE (p=0.01). Among the 276 patients without atherosclerosis observed for 1423 person-years, 7 VTE were observed compared to 2.6 expected (RR=2.7, 95%CI: 1.1, 5.6). Among the 291 patients with atherosclerosis (plaque ≥2mm), 4 VTE were observed compared to 5.8 expected (RR=0.7, 95%CI: 0.2, 1.8). Conclusions: Atherosclerosis is not a risk factor for incident VTE.

Abstract If the incidence of VTE varies by season, then the etiology of VTE could be related to seasonal exposures. However, the few studies that estimated VTE incidence by season reached conflicting conclusions. Objective: To estimate the incidence of VTE by month and by season of the year. Methods: Using the resources of the Rochester Epidemiology Project, we identified the inception cohort of Olmsted County, MN, residents with a first lifetime VTE over the 30-year period, 1966–1995 (n=2761). For each case, we reviewed the complete medical records in the community for the date of VTE onset, date of birth, gender and 48 baseline clinical characteristics commonly accepted as risk factors for VTE. We categorized cases as idiopathic (n=305) if no such characteristics were present; remaining cases were categorized as secondary. Incidence rates were calculated using the number of overall, idiopathic or secondary VTE cases by month of onset as the numerator, and age- and sex-specific estimates of the monthly population of Olmsted County as the denominator (yearly population estimates divided by 12). Results: The incidence of VTE did not vary appreciably by month of the year for overall, idiopathic or secondary VTE (all p-values >0.2). However, the overall incidence of VTE was higher in the 3 summer months (June, July, August) compared to the 3 winter months (December, January, February; 132.6 vs. 116.9 per 100,000 person-years, respectively, both age- and sex-adjusted to year 2000 U.S. Whites; p-value = 0.026). The same general pattern was true for both idiopathic and secondary VTE, but separately neither was statistically significant (p-value=0.48 and 0.15, respectively). Conclusions: The incidence of VTE is higher in the summer compared to the winter months, suggesting the hypothesis that vector-borne (e.g., mosquito, tick) and/or enteroviral infection may play a role in the etiology of VTE.

Abstract Abstract 1242 Background: While bivariate interactions between Factor V Leiden, prothrombin G20201A and hereditary deficiency of antithrombin, protein C and protein S compound VTE risk, whether other gene-gene interactions are associated with VTE is largely unknown. Objective: To test novel gene-gene interactions for an association with VTE. Methods: Genome-wide scan (Illumina 660Q [557,112 SNPs]) and candidate gene (12,551 SNPs in 764 genes within the anticoagulant, procoagulant, fibrinolytic and innate immunity pathways) genotypes from 1270 non-Hispanic adults of European ancestry with objectively-diagnosed VTE (cases; no cancer, venous catheter or antiphospholipid antibodies) and 1302 controls (frequency-matched on case age, gender, race, MI/stroke status) were merged and imputed to about 2.5 million SNPs with MACH using HapMap Phase 2 (60 CEU). We tested all pairwise SNP-SNP interactions using a 2-stage procedure (fast epistasis in the first stage and age, gender, MI/stroke status and state of residence adjusted analysis using logistic regression in the second stage). Results: The mean ± standard deviation case and control ages were 54.4 ± 16.2 and 55.6 ± 15.8 years, respectively and 51% were female. Among 2.299×10^12 pairwise interactions tested, 900 million reached a p-value<E-06 and were be tested in the second stage. Of these, 2417, 278 and 22 SNP-SNP interactions reached p-values ≤1E-9, 1E–10, and1E-11 respectively; 121 SNP-SNP interactions reached the Bonferroni statistical significance threshold (≤5.555659E-11). The gene-gene pairwise interactions with p-values≤1E-11 were: (1) chromosome 3 open reading frame 70 (C3orf70) – short-chain dehydrogenase/reductase (SDR family) member 4 (DHRS4), (2) 1-acylglycerol-3-phosphate O-acyltransferase 4 (AGPAT4) - NIMA (never in mitosis gene a)-related kinase 6 (NEK6), and (3) TatD DNase domain containing 2 (TATDN2) – WAS protein family, member 3 (WASF3). No gene function data are available for C3orf70. DHRS4 encodes for a 3-ketosteroid reductase important for active steroid hormone regulation, AGPAT4 is important for signal transduction and lipid biosynthesis, NEK6 encodes for Nek6 which downregulates p53-induced cancer cell senescence, TATDN2 encodes a deoxyribonuclease, WASF3 encodes for a member of the Wiskott-Aldrich syndrome protein family which are key regulators of signal transduction and cytoskeletal reorganization in hematopoietic cells. Conclusion: Pairwise gene-gene interaction analyses suggest potential associations between VTE and novel genes for future testing in replication studies. Disclosures: Heit: Daiichi Sankyo: Consultancy, Honoraria.

Abstract Abstract 2295 Background: While interaction between Factor V Leiden and other VTE risk exposures compound VTE risk, whether other gene-environment interactions are associated with VTE is largely unknown. Objective: To test novel gene-environment interactions for an association with VTE. Methods: Genome-wide scan (Illumina 660Q [557,112 SNPs]) and candidate gene (12,551 SNPs in 764 genes within the anticoagulant, procoagulant, fibrinolytic and innate immunity pathways) genotypes from 1270 non-Hispanic adults of European ancestry with objectively-diagnosed VTE (cases; no cancer, venous catheter or antiphospholipid antibodies) and 1302 controls (frequency-matched on case age, gender, race, MI/stroke status) were merged and imputed to 2.5 million SNPs with MACH using HapMap Phase 2 (60 CEU). We tested pairwise interactions between these SNPS and each of 7 exposures (“stress” [defined as ever hospitalized, ever surgery, ever trauma, ever transfemoral procedure or ever leg paresis], ever surgery, ever neurosurgery, ever orthopedic surgery; and among women, ever pregnant, ≥3 pregnancies and ever oral contraceptives/hormone therapy), adjusted for age, gender, MI/stroke status and state of residence, using logistic regression. Results: The mean ± standard deviation case and control ages were 54.4 ± 16.2 and 55.6 ± 15.8 years, respectively and 51% were female. Among the 7 models, a significant interaction was found between stress and A2BP1 (rs735919, OR=0.2, p=2.8E-07), ever surgery and ABCG8 (rs6709904, OR=0.27, p=2.1E-07), orthopedic surgery and SLC24A3 (rs6081599, OR=0.53, p=2.8E-07), neurosurgery and TBC1D4 (rs17064485, OR=8.5, p=3.5E-07), every pregnant and MICALCL (rs11022321, OR=0.17, p=8.3E-07), >3 pregnancies and KLRB1 (rs2536929, OR=0.31, p=1.1E-07) and ever oral contraceptives/hormone therapy and LINGO2 (rs10969259, OR=2.02, p=4.9E-07). A2BP1 plays a role in RBC shape and size, ABCG8 is associated with severe hypercholesterolemia, SLC24A3 is important for intracellular calcium homeostasis, TBC1D4 participates in cell signaling in CD4+/CD45RO+ memory T-cells, MICALCL is an extracellular signal-regulated kinase 2-binding protein, KLRB1 regulates natural killer cell function, and LINGO2 encodes a leucine-rich repeat signaling protein. Our observed interaction of VTE risk factors with these genes to compound the risk of VTE requires replication in future studies. If confirmed, functional studies will be important to determine the roles of these gene products in the causal pathway to VTE. Conclusion: VTE risk factors and SNPs within several novel genes may interact to compound the risk of VTE. Disclosures: Heit: Daiichi Sankyo: Consultancy, Honoraria.

Abstract Background Recurrent VTE is frequent and may be fatal; the 2-week case fatality rate after recurrent pulmonary embolism is 11%. However, predictors of VTE recurrence remain uncertain. Objective To identify independent predictors of VTE recurrence, including interim exposures, adjusted for baseline characteristics and treatments. Methods Using the resources of the Rochester Epidemiology Project, we identified all Olmsted County, MN residents with objectively-diagnosed incident VTE over the 13-year period, 1988-2000. All patients surviving ≥1 day were followed forward in time for first objectively-diagnosed VTE recurrence, death or other loss to follow-up, or 12/31/2005, whichever came first. VTE recurrence was defined as new thrombosis of a previously uninvolved venous territory or re-thrombosis of a previously involved venous territory after interim clearing of the incident thrombosis. We estimated the cumulative incidence of VTE recurrence using the Kaplan Meier product limit method, both overall and by incident cancer-associated, idiopathic and non-cancer secondary VTE. For all patients with VTE recurrence, and a random sample of all patients with incident VTE, we reviewed the complete medical records in the community and collected demographic and baseline characteristics, treatments (heparin and warfarin [including proportion of time in therapeutic range], IVC filter, aspirin, statins) and interim VTE risk factor exposures. Using a case-cohort study design with Barlow’s weighting scheme, we tested these demographic, baseline and interim characteristics as potential predictors of VTE recurrence using time-dependent Cox proportional hazards modeling. Results Among 1261 incident VTE patients who survived ≥1 day, 306 developed recurrent VTE over 6,485 person-years of follow-up. The cumulative incidence of VTE recurrence at 30 and 180 days, and 1 and 5 years, were 4.5%, 8.2%, 10.5% and 20.3%, respectively. The five-year cumulative recurrence rates among patients with incident cancer-associated, idiopathic and non-cancer secondary VTE were 42.6%, 26.8% and 18.5%, respectively. Among the random cohort of 415 incident VTE cases, 381 (92%) and 359 (87%) received heparin and warfarin therapy for a median (IQR) 5 (4 - 7) and 119 (63 - 194) days, respectively. From multivariable analysis, interim exposure to trauma, major surgery (general, cardiac or gynecologic), infection, hospitalization for acute medical illness or active cancer after the incident VTE were each associated with an increased hazard of VTE recurrence, while use of warfarin and achievement of a therapeutic APTT within the first day of starting heparin were associated with reduced hazards. Although measures of time in therapeutic range while on heparin or warfarin were associated with a reduced risk of recurrence from univariate analyses, these results were not significant when adjusting for other risk factors. Conclusions Among patients with incident VTE, new exposure to trauma, major surgery, infection, hospitalization for acute medical illness or active cancer are independent predictors of VTE recurrence; patients with such exposures should be considered for VTE prophylaxis. Disclosures: No relevant conflicts of interest to declare.

Abstract Abstract 480 Background: While interaction between Factor V Leiden and other VTE risk exposures (i.e., oral contraceptives, hormone therapy, pregnancy, cancer, minor trauma) compound VTE risk, whether other gene-environment interactions are associated with VTE is largely unknown. Objective: To test gene-environment interactions for an association with VTE. Methods: Cases (n=1488) were Mayo Clinic European-American patients of non-Hispanic ancestry with objectively-diagnosed VTE in the absence of active cancer, venous catheter or antiphospholipid antibodies. Controls (n=1439) were Mayo Clinic outpatients without VTE who were frequency-matched on case age, gender, race, MI/stroke status and state of residence. We selected candidate genes relevant to the anticoagulant, procoagulant, fibrinolytic and innate immunity pathways, focusing on platelet, monocyte, neutrophil and endothelial cell agonists, receptors, ligands, signal transduction and adhesion molecules, granule contents and effectors; plasma proteases and inhibitors; matrix metalloproteases; inflammatory cytokines and receptors; estrogen, progesterone and androgen receptors; co-regulators and enzymes related to estrogen metabolism; important enzymes for catechol, homocysteine, thromboxane A2 and prostacyclin biosynthesis and metabolism; and HMG CoA reductase. For these genes (n=750), we selected all non-synonymous coding single nucleotide polymorphisms (SNPs) with minor allele frequency ≥0.5%; the remaining SNPs were selected using an LD tagging algorithm (Carlson et al. AJHG 2004); 479 ancestry-informative markers were also included (total n=13,241 SNPs). Leukocyte genomic DNA was genotyped using a custom Illumina Infinium iSelect platform, including appropriate controls. We tested all pairwise interactions between 12,483 SNPs (12296 autosomal, 187 chromosome X) that passed quality control and each of seven exposures (ever “stress”[defined as ever hospitalized, ever surgery, ever trauma, ever transfemoral procedure or ever leg paresis], ever surgery, ever neurosurgery, ever orthopedic surgery; and among women, ever pregnant, ≥3 pregnancies and ever oral contraceptives/hormone therapy), adjusted for age, gender, MI/stroke status and state of residence, using logistic regression. False discovery rates (q-value) were calculated to estimate the expected fraction of false positive associations due to multiple statistical tests. Results: The mean ± SD case and control ages were 55.3 ± 16.4 and 56.5 ± 15.9 years, respectively, and 51% were female. Analysis of ancestry-informative markers revealed no evidence of population stratification. Among the seven models, a significant interaction (q≤0.05) was found between ever surgery and two genes located on chromosome 1: USF1 (rs2516840, OR=2.1, p=2.0E-06; q=0.02), and F11R (rs790055 & rs790056; OR∼2.29, p∼4E-06; q=0.02); the two F11R SNPs are in complete linkage dysequilibrium (LD), and the USF1 SNP is in moderate LD with the F11R SNPs (r2∼0.6). USF1 encodes for upstream transcription factor 1 which regulates many genes involved in lipid and glucose homeostasis. F11R encodes for the platelet F11 receptor (JAM [junctional adhesion molecule], a cell adhesion molecule important for platelet adhesion to cytokine-stimulated endothelial cells. Conclusion: Interaction between surgery and SNPs within USF1 and F11R are associated with significantly increased risks for VTE. These potential associations require confirmation in future replication studies. Disclosures: No relevant conflicts of interest to declare.

Abstract Abstract 476 Background: The burden of VTE among nursing home (NH) residents is known to be high. Yet very little data exist to help determine which NH residents to target for VTE prophylaxis. The need to characterize VTE risk in this population is especially great because the advanced age and high comorbidity that typify NH residents contribute to increased likelihood of adverse consequences from VTE prophylaxis. To compound the problem, results of our prior univariate analyses of certain factors known to contribute substantially to VTE risk in the general population (e.g., major surgery, medical hospitalization, trauma) suggest that impact of these factors on VTE may be less substantial among NH residents. Objective: To use the longitudinal population-based resources of the Rochester Epidemiology Project (REP) to investigate the contribution of multiple clinical characteristics to risk of VTE among NH residents. Methods: We took advantage of the previous identification of all Olmsted County, MN residents who met research criteria for incident VTE 1988 through 2005 (N=2,332). We then determined which individuals were resident of a local NH at time of VTE symptom onset, regardless of location of symptom onset (i.e., individuals whose VTE occurred in-hospital having been admitted from a NH were considered NH residents). For each such NH VTE case (N=269), we identified 2 same sex Olmsted County residents of similar age and duration of medical history who were also resident of a local NH at the time of the case's VTE event (i.e., index date) (N=538). We reviewed the detailed provider-linked medical records of NH VTE cases and NH non-VTE controls for 3 months before index for information on multiple characteristics identified or hypothesized as contributing to VTE risk in studies of the general population and other at-risk subgroups by our group and others (e.g., patient demographics, body mass index, major surgery, hospitalization for acute medical illness, outpatient surgical procedures, trauma/fracture, leg paresis, active cancer, superficial vein thrombosis, varicose veins, infections, diabetes mellitus, coronary artery disease, congestive heart failure, and multiple medications, including anticoagulants, statins, beta-blockers, ACE inhibitors and angiotensin II receptor antagonists). We tested and estimated the odds ratio associated with each factor using step-wise conditional logistic regression. Variables for which <10 cases or <10 controls exhibited the characteristic were excluded from analysis. Results: The first five variables to enter the model were urinary tract infection, active cancer, superficial vein thrombosis, pneumonia, and leg paresis. The respective odds ratios (95% confidence intervals) with all five in the model were 1.7 (1.2, 2.4); 2.1 (1.3, 3.5); 2.1 (1.3, 3.4); 1.9 (1.3, 2.8); 2.3 (1.3, 4.2); each p value was <0.01. Variables associated with high VTE risk in the general population (i.e., surgery, hospitalization for medical illness, trauma/fracture) were not included in the top five risk factors for NH residents. Conclusions: Infection is a potent VTE risk factor among NH residents. Our study results will help inform development of practice guidelines in the NH and stimulate future research on putative VTE mechanisms. REP data afford limited information on cognitive and physical disability, immobility, and need for NH care. Additional investigations are needed that combine information from clinical and NH assessments. Disclosures: No relevant conflicts of interest to declare.