Abstract
Background
Independent risk factors for cancer-associated incident venous thromboembolism (VTE) and their magnitude of risk are not fully characterized.
Aim
To identify non-cancer and cancer-specific risk factors for cancer-associated incident VTE.
Methods
In a population-based retrospective case-control study, we used Rochester Epidemiology Project and Mayo Clinic Cancer Registry resources to identify all Olmsted County, MN residents with active cancer-associated incident VTE, 1973-2000 (cases; n=570) and 1-3 residents with active cancer matched to each case on age, sex, date and duration of active cancer (controls; n=604). Using conditional logistic regression, we tested cancer and non-cancer characteristics for an association with VTE, including a cancer site VTE risk score.
Results
In the multivariable model, higher cancer site VTE risk score (OR=1.4 per 2-fold increase), cancer stage ≥2 (OR=2.2), liver metastasis (OR=2.7), chemotherapy (OR=1.8) and progesterone use (OR=2.1) were independently associated with VTE, as were BMI<18.5 kg/m2 (OR=1.9) or ≥35 kg/m2 (OR=4.0), hospitalization (OR=7.9), nursing home confinement (OR=4.7), central venous (CV) catheter (OR=8.5) and any recent infection (OR=1.7). In a subgroup analysis, platelet count ≥350×109/L at time of cancer diagnosis was marginally associated with VTE (OR=2.3, p=0.07).
Conclusion
Cancer site, cancer stage ≥2, liver metastasis, chemotherapy, progesterone, being underweight or obese, hospitalization/nursing home confinement, CV catheter, and infection are independent risk factors for incident VTE in active cancer patients.
Keywords: Venous thromboembolism, cancer, epidemiology, risk factors, deep vein thrombosis, pulmonary embolism
Introduction
Active cancer is an independent risk factor for VTE, with an overall 4- to 7-fold increased risk [1-7], and accounts for 20-30% of all new VTE events in the community.[8-11] VTE commonly complicates cancer and is associated with significant morbidity.[12-14] Acute VTE may delay or modify anti-cancer therapy, and usually leads to chronic anticoagulation. Compared to VTE patients without cancer, individuals with cancer and VTE have a higher risk for hemorrhage with anticoagulant therapy [15-21] such that primary or secondary pharmacologic VTE prophylaxis for all active cancer patients may be inappropriate. Furthermore, cancer patients with VTE have worse survival than cancer patients without VTE.[9, 14, 20, 22-28]
While the risk of VTE appears to vary by cancer site [6, 7, 15, 23, 29-34], cancer stage and grade [5, 6, 23, 35], chemotherapy use [2, 31, 36-39], and patient-related characteristics [5, 40-45], whether these cancer- and non-cancer characteristics are independent risk factors for VTE and the magnitude of the associated risk are uncertain and not fully characterized. To address this important gap in knowledge, we conducted a population-based case-control study of active cancer patients to test cancer and non-cancer characteristics as potential risk factors for incident VTE, as these data may allow identifying cancer patients at a high risk for developing incident VTE.
Methods
Study Setting and Design
Using the longitudinal and population-based resources of the Rochester Epidemiology Project (REP) [46], we identified all Olmsted County, MN residents with incident deep vein thrombosis (DVT) and/or pulmonary embolism (PE), over the 35-year period, 1966-2000, as previously described.[47, 48] For this study, we included all Olmsted County residents who consented to access of their medical records for research purposes and had a first lifetime symptomatic leg DVT or PE during the 28-year period, 1973-2000, and active cancer in the three months prior to or after the VTE diagnosis as previously defined.[28, 49] Briefly, a cancer was considered as active if any evidence of cancer (i.e., diagnosis, treatment, progression, oncologist review) was found via medical record review within 92 days prior to the VTE event/index date. At a minimum, any newly diagnosed cancer was considered as active for 6 months from the diagnosis date if not treated by curative surgery. Patients undergoing curative surgery were considered cancer-free if the surgical margins were clear and no cancer recurred within three months after surgery. Questions regarding curative surgery were adjudicated by a committee of the investigators (AAA, JAH, RSM) who had access to all medical records. Multiple myeloma, myelodysplastic syndrome and myeloproliferative neoplasms (MPN) were considered as permanently active. Lymphoma, chronic lymphocytic leukemia (CLL) and prostate cancer were considered as active for as long as they were left untreated (assumed to reflect a planned observation of the cancer), and any stage 4 cancer was considered as permanently active (with the exception of stage 4 Hodgkin’s lymphoma, intermediate/high grade non-Hodgkin’s lymphoma, and testicular or germ cell cancer). With the exception of CLL, any cancer classified as stage 0 was not categorized as active until there was stage progression. Objectively diagnosed symptomatic VTE and those diagnosed by a physician where the level of suspicion was sufficiently high that the patient was treated by anticoagulants were included. We included the latter because physicians are hesitant to use anticoagulants in cancer patients due to the high bleeding risk such that any use of anticoagulant therapy would indicate a high suspicion of VTE. Patients with incidental VTE, jugular or isolated arm DVT, or sub-segmental PE were excluded.
Using the resources of the Mayo Clinic Cancer Registry and the REP, we identified all Olmsted County residents with first lifetime active cancer and no prior VTE whose cancer was diagnosed within ± 5 years of the case’s cancer diagnosis date.[2] The Cancer Registry maintains patient demographics (including age, sex, race/ethnicity and residence at time of diagnosis), cancer information (i.e., primary site, histology and extent of disease), and cancer treatment information (i.e., surgery, radiation therapy, chemotherapy, hormone, immunotherapy, or other types of treatment). We further matched on sex, date of birth (± 5years), date of first Olmsted County medical record (± 5years), and duration of active cancer (i.e., the control’s active cancer duration was ≥ the case’s duration to ensure that the control would have equal exposure to the risk of VTE induced by cancer). From this, a random sample of potential controls was identified. The nurse abstractors confirmed no prior history of VTE and no VTE within 3 months post-index, and that control’s active cancer duration was equal to or up to two years greater than the case’s active cancer duration. The control index date was the first medical visit within this 2 year window.
For cases with more than one primary active cancer at the time of the incident VTE, a separate control was matched (matched on each primary active cancer) using the same criteria as above. For those individuals with multiple primary active cancers of the same type (breast cancer in both breasts, lung cancer in different lobes, etc.), we kept the case-control pair where the case had the higher stage. The study was approved by the Mayo Clinic and Olmsted Medical Center Institutional Review Boards.
Measurements
Using explicit criteria, trained and experienced nurse abstractors reviewed all medical records in the community and collected data on date and type of incident VTE, baseline characteristics (age at incident VTE; sex; hospitalization with or without surgery [categorized as general, neurological, orthopedic, cardiac, or gynecologic surgery], anesthesia use, nursing home confinement, trauma/fracture, immobilization, infection and central venous catheter/ pacemaker, all within the 92 days prior to the VTE event/index date; any prior history of congestive heart failure, neurologic disease with leg paresis, chronic renal disease, liver disease, chronic lung disease, heart disease, diabetes, hyperlipidemia, autoimmune connective tissue disease, superficial vein thrombosis and thoracic outlet syndrome), and vital status at last clinical contact, as previously performed.[2, 28, 50-52] Body mass index was based on the nearest height and weight in the two years prior to the VTE event/index date, if available, or within three months after the event date. Additional data were collected from the Mayo Clinic Cancer Registry and medical records on the date of diagnosis of active cancer, cancer site and stage, tumor-node-metastasis (TNM) classification and cancer grade at the time of cancer diagnosis; cancer histopathology; presence of venous invasion by cancer, lymph node involvement, and metastasis to liver and/or bone; hormonal therapy, chemotherapy, and radiation therapy. We categorized chemotherapeutic agents into the following categories: a) Alkylators (e.g., cyclophosphamide, ifosfamide, busulfan, carmustine, melphalan, procarbazine, dacarbazine); b) Antimetabolites (e.g., cytarabine, 5-fluoro-uracil, methotrexate, hydroxyurea, gemcitabine); c) Anthracyclines (e.g., doxorubicin, daunorubicin, idarubicin, mitoxantrone); d) Alkaloids (e.g., etoposide, irinotecan, topotecan, vincristine, vinorelbine); e) Platinum (e.g., cisplatin, carboplatin, oxaliplatin); f) Hormonal modulators (e.g., goserelin, leuproloid); g) Immunomodulators (e.g., thalidomide, lenalidomide, interferon); and h) Other (e.g., imatinib, rituximab, L-asparaginase), and evaluated whether one subclass of chemotherapy was more strongly associated with VTE than others. In a subset of cases and controls diagnosed with active cancer between 1983 -2000, for whom laboratory data were available in the Mayo Clinic’s electronic medical records, and who had a complete blood count performed within 14 days prior to or four days after the cancer diagnosis date, the hemoglobin, leukocyte count and platelet count closest to the date of cancer diagnosis were captured electronically. The study oncologists (AAA, RSM) also re-staged the cancer(s) at the time of the incident VTE or the control’s index date. Cancer characteristics specific to the cancer (cancer site and stage) were recorded as present only if documented within the 92 days prior to the VTE event/index date, or, for cases only, within 92 days after the incident VTE event; chemotherapy, radiation therapy and hormone therapy characteristics were recorded as present if documented within the 92 days prior to the index event. All other cancer characteristics were recorded as present if documented any time prior to the index event. All cancer stage assignments were adjudicated by the study oncologists and based on the American Joint Committee on Cancer staging guidelines (6th edition).[53]
Analysis
All modeling was done using conditional logistic regression, which stratifies on the matched set. Because two controls could be matched on a single case but by differing criteria (differing duration since the malignancy diagnosis for each malignancy) we used the robust sandwich estimator of Lin and Wei [54] for the covariance matrix to account for clustering within the strata. The individual sets of observations (1 case: 1-3 controls) were assumed to be independent.
Rather than arbitrarily choosing one cancer site as a comparison group in our modeling, we evaluated the association of cancer site and VTE using standardized morbidity ratios (SMR; “cancer site VTE risk score”) based on the ratio of observed cancers to expected cancers in each of 26 possible cancer sites among incident VTE cases, 1988-2000, as previously performed.[34] In brief, we identified all active cancers by cancer site among Olmsted County residents with incident VTE, 1988-2000. We used Iowa State Surveillance and Epidemiology and End Results (SEER) data to estimate the expected age- and sex-specific prevalence of cancer by cancer site among incident VTE cases, and calculated SMR for each cancer site by dividing the observed by the expected number of cancers.[34] Using the SMR as a “cancer site VTE risk score” had the advantage of using a continuous variable instead of 26 dichotomous variables and it jointly adjusted for cancer location and the magnitude of risk, giving greater weight to cancers with a higher risk for VTE. In turn this allowed us to focus more closely on cancer related variables other than cancer site. Because cancer site and stage are highly confounded, we adjusted for stage before the other cancer-related variables, choosing the combination of stages 2–4 that had the strongest association with incident VTE. Further, the cancer site VTE risk score was based on cancer prevalence, not incidence; we verified and adapted the score fit in the current context by first adjusting for cancer stage, then introducing into the model each cancer site in turn, one at a time. Cancer sites which showed a significant residual association with VTE were left in the model as these represented sites where the cancer site VTE risk score did not adequately fit the data. We then adjusted the multivariable conditional logistic model for the cancer site VTE risk score and the additional significant cancer site and stage variables. The remaining cancer-specific variables and the known non-cancer independent VTE risk factors were tested for an association with VTE using a forward stepwise selection with a p-value ≤ 0.05 to stay in the model. We tested the variables in the final model for interactions with the matching variables (sex, age, time since cancer, calendar year) and for interactions among all main effect variables in the final model. The cancer site VTE risk score was highly skewed so we used the natural log in all modeling. Because the remaining cancers in the score comparison group had a score of 0, we added 1.0 to all scores before taking the natural log.
Because pre-chemotherapy platelet count ≥350 × 109/L, hemoglobin <100g/L, and leukocyte count >11 × 109/L have been shown to be risk factors for cancer-associated VTE [40, 43], we explored the association of these covariates with VTE univariately (after adjusting for the cancer site VTE risk score, additional cancer sites, and stage), and after adjusting for the final model determined above, in the subset of cases and controls where these data were available. Furthermore, as clinical practice in cancer diagnosis and therapy have evolved over time with introduction of new diagnostic procedures and therapeutic agents and/or combination chemotherapy regimens, we formally tested for interaction between VTE event year and the variables included in the final multivariable model. In addition, we estimated the risk of VTE over two time frames which represented milestones in the introduction of new diagnostic procedures and therapeutic agents (1973-1985 [introduction of alkylators, nucleoside analogues, antimetabolites, and anthracycline] and 1986-2000 [introduction of computed tomography (CT) for staging; taxanes, platinum-based antineoplastic agents, hematopoietic growth factors, and targeted therapies such as rituximab; increase in outpatient cancer management]) to evaluate whether the association of VTE with cancer stage, chemotherapy and location of onset of VTE changed in these time frames.
Results
Over the 28-year period, 1973-2000, 2782 Olmsted County residents were diagnosed with incident VTE (DVT alone, n=1344; PE with or without [±] DVT, n=1432; chronic thromboembolic pulmonary hypertension, n=6), of whom 570 (20.5%) had active cancer; 31 (5%) had more than one primary cancer site (28 residents had two primary cancers and 3 residents had three primary cancers for a total of 34 additional cancers). Nineteen (3.3%) had their underlying cancer diagnosed at autopsy, of whom 13 (68.4%) had their VTE diagnosed concomitantly at autopsy and the remaining six (31.6%) had their VTE event diagnosed within two weeks prior to autopsy date; 190 (33.3%) had their incident VTE event within 3 months after the cancer diagnosis date; and 61 (10.7% of 570 people) had their VTE event within three months prior to the cancer diagnosis. Eighty seven percent (n=495) of VTE were objectively diagnosed; 13% (n=75) were clinically diagnosed VTE where the level of suspicion was sufficiently high that these subjects were treated with anticoagulants. The distribution of VTE by event type was DVT alone (n=254) and PE ± DVT (n=316); 143 (45.2%) of all the PE’s were discovered on autopsy (63 in 1973-1980; 49 in 1981-1990; 31 in 1991-2000). The mean (SD) and median (IQR) patient age at the incident VTE event date were 68.5 (14.7) and 70.9 (60.0 - 79.5) years, respectively; 49% were female. The mean (SD) control age and sex were 69.2 (14.0) years and 49% females, respectively. The mean (SD) and median (IQR) person-years of follow-up were 36.0 (20.3) and 37.0 (20 - 51) for cases; and 36.9 (20.0) and 38.0 (21 - 52) for controls respectively. Most VTE events occurred near the time of cancer diagnosis (median: 142.5 days; IQR: 19 - 592).
The cancer and non-cancer baseline characteristics and prevalence of VTE risk factors among VTE cases and controls are shown in Tables 1 and 2. Compared to controls, VTE cases had a higher cancer site VTE risk score; more stage 4 disease and liver and bone metastases; higher cancer grade; were more likely to have received chemotherapy, radiation therapy and progestins; more frequent cancer stage progression in the time interval between cancer diagnosis and VTE event/ index date; were more often underweight or markedly obese; underwent surgery or received anesthesia more frequently; were hospitalized or in a nursing home more frequently; and had a higher prevalence of neurological disease with leg paresis, renal disease, heart disease, autoimmune connective tissue disease, immobilization, trauma/fracture, central vein catheter, prior superficial vein thrombosis and infection (Tables 1 and 3b).
Table 1. Baseline Characteristics of Patients with Active Cancer and Venous Thromboembolism - Cases and Controls.
Baseline Characteristic | Cases (n=570) |
Controls (n=604) |
---|---|---|
Age, mean ± SD | 68.5 (14.7) | 69.2 (14.0) |
Female, n (%) | 279 (49.0) | 296 (49.0) |
Body mass index (BMI, kg/m2), mean ± SD | 25.3 (6.5) | 25.5 (5.1) |
BMI <18.5 kg/m2 (Underweight), n (%) | 66 (11.6) | 37 (6.1) |
BMI 18.5-25 kg/m2 (Normal), n (%) | 225 (39.5) | 253 (41.9) |
BMI 25-30 kg/m2 (Overweight), n (%) | 175 (30.7) | 220 (36.4) |
BMI 30-35 kg/m2 (Obese class I), n (%) | 67 (11.8) | 68 (11.3) |
BMI >35 kg/m2 (Obese class II-III), n (%) | 37 (6.5) | 26 (4.3) |
Cancer site VTE risk score*†, median (inter quartile range) |
12.6 (7.9, 19.8) | 8.6 (7.9, 13.9) |
Stage*, n (%) | ||
0 | 7 (1.2) | 26 (4.3) |
1 | 137 (22.7) | 214 (35.5) |
2 | 117 (19.4) | 120 (19.9) |
3 | 122 (20.2) | 102 (16.9) |
4 | 221 (36.6) | 140 (23.3) |
Stage progression*, n (%) | 148 (24.7) | 100 (16.6) |
Metastasis to liver*, n (%) | 141 (23.5) | 48 (8.0) |
Metastasis to bone*, n (%) | 97 (16.1) | 66 (11.0) |
Tumor grade 3 or 4*, n (%) | 394 (66.2) | 321 (55.3) |
Chemotherapy‡, n (%) | 166 (29.2) | 108 (17.9) |
Alkylators§ | 67 (11.8) | 59 (9.8) |
Antimetabolites§ | 78 (13.7) | 37 (6.2) |
Anthracyclines§ | 44 (7.8) | 21 (3.5) |
Alkaloids§ | 33 (5.8) | 20 (3.3) |
Platinum§ | 35 (6.2) | 20 (3.3) |
Hormonal modulators§ | 0 (0.0) | 3 (0.5) |
Immunomodulators§ | 4 (0.7) | 3 (0.5) |
Other§ | 73 (12.8) | 54 (9.0) |
Radiation therapy‡, n (%) | 54 (9.5) | 32 (5.3) |
Tamoxifen‡, n (%) | 24 (4.2) | 19 (3.2) |
Estrogen/ Oral Contraceptives‡, n (%) | 26 (4.6) | 27 (4.5) |
Women only, n (%) | 18 (6.5) | 24 (8.1) |
Progestins‡, n (%) | 44 (7.7) | 28 (4.6) |
Women only, n (%) | 27 (9.7) | 24 (8.1) |
Diethyl stilbesterol‡, n (%) | 17 (3.0) | 18 (3.0) |
Trauma/fracture‡, n (%) | 32 (5.6) | 16 (2.6) |
Hospitalized∥, n (%) | 392 (68.8) | 206 (34.1) |
ICU stay, n (%) | 78 (14.2) | 20 (3.4) |
General surgery, n (%) | 131 (23.0) | 98 (16.2) |
Neurological surgery, n (%) | 12 (2.1) | 2 (0.3) |
Orthopedic surgery, n (%) | 19 (3.3) | 4 (0.7) |
Gynecologic surgery¶, n (%) | 21 (7.5) | 12 (4.0) |
Any anesthesia‡, n (%) | 245 (43.0) | 163 (27.0) |
General anesthesia, n (%) | 181 (31.8) | 119 (19.7) |
Epidural and spinal anesthesia, n (%) | 64 (11.2) | 44 (7.3) |
Nursing home confinement‡, n (%) | 22 (3.9) | 20 (3.3) |
Neurological disease with leg paresis, n (%) | 40 (7.0) | 7 (1.2) |
Immobilization‡, n (%) | 60 (10.5) | 17 (2.8) |
Central venous catheter‡, n (%) | 66 (11.6) | 11 (1.8) |
Varicose veins, n (%) | 159 (27.9) | 145 (24.0) |
Prior superficial vein thrombosis, n (%) | 79 (13.9) | 49 (8.1) |
Diabetes, n (%) | 62 (11.0) | 71 (11.8) |
Hyperlipidemia, n (%) | 133 (23.5) | 157 (26.0) |
Lipid lowering drugs‡, n (%) | 19 (3.4) | 30 (5.0) |
Statins, n (%) | 10 (1.8) | 24 (4.2) |
Chronic renal disease, n (%) | 17 (3.0) | 5 (0.8) |
Autoimmune connective tissue disease#, n (%) | 16 (2.8) | 7 (1.2) |
Any heart disease, n (%) | 192 (33.7) | 173 (28.6) |
Lung disease/ pulmonary hypertension, n (%) | 130 (22.8) | 131 (21.7) |
Any infection‡, n (%) | 227 (39.8) | 100 (16.6) |
Ever Smoker, n (%) | 303 (53.2) | 341 (56.5) |
Anticoagulation use‡, n (%) | 24 (4.2) | 16 (2.6) |
Thirty-one of VTE cases had more than one cancer primary site (34 cancers), and each primary site was matched to a control, matching on age (± 5 years), sex, cancer diagnosis date (± 5 years), and duration of prior medical history (± 5 years). The cancer specific characteristics are based on the 604 total malignancies as these elements were collected for each of malignancy diagnoses.
The Cancer site VTE risk score is the cancer site standardized morbidity ratio for VTE (see Table 2. [34])
Within 92 days prior to the VTE event/ index date.
The subclasses of chemotherapy do not add up to 100%, as subjects could be on combination chemotherapy.
In hospital or within 92 days of previous hospitalization compared to no hospitalization in the past 92 days. Includes patients hospitalized with surgery or for acute medical illness.
Among women only (279 cases, 296 controls)
includes rheumatoid arthritis, systemic lupus erythematosus, scleroderma
Table 2. Cancer Site and Standard Morbidity Ratio (SMR; “Cancer Site VTE Risk Score”)* Distribution of Patients with Active Cancer and Venous Thromboembolism: by Case and Control Status.
Cancer Site | Cases n (%) |
Controls n (%) |
Cancer site VTE risk score (SMR*) |
---|---|---|---|
Brain | 18 (3.0) | 14 (2.3) | 47.2 |
Pancreatic | 52 (8.6) | 13 (2.2) | 42.0 |
Other digestive | 18 (3.0) | 8 (1.3) | 30.9 |
Lymphoma | 41 (6.8) | 38 (6.3) | 28.8 |
Leukemia | 20 (3.3) | 34 (5.6) | 19.8 |
Stomach | 18 (3.0) | 7(1.2) | 18.2 |
Multiple myeloma | 8 (1.3) | 15 (2.5) | 14.0 |
Kidney | 11 (1.8) | 17 (2.8) | 13.9 |
Lung | 72 (11.9) | 60 (9.9) | 13.0 |
Ovary† | 17 (5.7) | 16 (5.4) | 13.0 |
Myeloproliferative neoplasm | 19 (3.2) | 1 (0.2) | 12.6 |
Bladder | 27 (4.5) | 19 (3.2) | 11.0 |
Melanoma | 10 (1.7) | 11 (1.8) | 9.7 |
Breast‡ | 61 (20.6) | 100 (33.8) | 8.6 |
Other gynecologic‡ | 24 (8.1) | 22 (7.4) | 8.4 |
Prostate§ | 71 (23.0) | 104 (33.8) | 7.9 |
Colon/rectal | 75 (12.4) | 74 (12.2) | 7.3 |
Other cancers combined‡ | 36 (6.0) | 51 (8.4) | ------ |
Standard morbidity ratio (SMR; “Cancer site VTE risk score”) among Olmsted County, MN residents with incident venous thromboembolism, 1988-2000; adapted from Petterson, T.M., et al. [34]
Women only (296 cases, 296 controls);
Men only (308 cases, 308 controls).
Remaining cancers combined (head and neck, liver, other thorax, bone, soft tissue, myelodysplastic syndrome, other genitourinary, eye, misc. cancer [other] with scores of 4.1, 25.9, 0, 0, 5.0, 7.0, 10.6, 26.8, and 9.1, respectively).
Table 3b. Univariate Analysis for Predictors of Venous Thromboembolism in Cancer Patients, Adjusting for Cancer Site VTE Risk Score, Individual Cancer Sites and Cancer Stage (Table 3a).
Characteristic | OR | 95% CI* | p-value* | |
---|---|---|---|---|
| ||||
Stage change | 1.99 | 1.39 | 2.85 | <.001 |
| ||||
Metastasis | 1.70 | 1.27 | 2.29 | <.001 |
| ||||
Metastasis to liver | 2.62 | 1.81 | 3.79 | <.001 |
| ||||
Metastasis to bone | 1.56 | 1.05 | 2.32 | 0.027 |
| ||||
Cancer grade 3 or 4 | 1.38 | 1.04 | 1.82 | 0.026 |
| ||||
Chemotherapy | 1.76 | 1.27 | 2.45 | <.001 |
| ||||
Radiation therapy | 1.77 | 1.00 | 3.15 | 0.051 |
| ||||
Progestins | 1.76 | 1.03 | 2.99 | 0.038 |
| ||||
Age, 10-year increase | 0.67 | 0.43 | 1.06 | 0.084 |
| ||||
BMI (continuous variable) | 0.99 | 0.97 | 1.02 | 0.67 |
| ||||
BMI (categorical; Reference: 18.5-25 kg/m2) | ||||
BMI <18.5 kg/m2 (Underweight) | 1.93 | 1.18 | 3.17 | 0.009 |
| ||||
BMI 25-30 kg/m2 (Overweight) | 0.87 | 0.64 | 1.18 | 0.38 |
| ||||
BMI 30-35 kg/m2 (Obese class I) | 1.13 | 0.74 | 1.72 | 0.58 |
| ||||
BMI >35 kg/m2 (Obese class II-III) | 1.70 | 0.89 | 3.25 | 0.11 |
| ||||
Location of onset of VTE (Reference: Community) | ||||
Hospitalized | 10.87 | 6.48 | 18.22 | <.001 |
| ||||
Hospitalization within 92 days | 4.94 | 3.37 | 7.24 | <.001 |
| ||||
Nursing Home/no hospitalization | 4.46 | 1.92 | 10.34 | <.001 |
| ||||
ICU stay | 5.26 | 2.79 | 9.91 | <.001 |
| ||||
General surgery | 2.28 | 1.57 | 3.31 | <.001 |
| ||||
Neurosurgery | 8.48 | 0.93 | 77.52 | 0.058 |
| ||||
Orthopedic surgery | 3.75 | 1.34 | 10.46 | 0.012 |
| ||||
Gynecologic surgery | 2.32 | 1.09 | 4.95 | 0.028 |
| ||||
General/cardiac/orthopedic/neuro/gynecologic surgery |
3.27 | 2.25 | 4.74 | <.001 |
| ||||
Any anesthesia | 2.91 | 2.09 | 4.06 | <.001 |
| ||||
Superficial vein thrombosis | 1.91 | 1.28 | 2.87 | 0.002 |
| ||||
Chronic renal disease | 2.92 | 0.92 | 9.30 | 0.069 |
| ||||
Neurologic disease | 6.45 | 2.49 | 16.69 | <.001 |
| ||||
Any heart disease | 1.41 | 1.04 | 1.90 | 0.026 |
| ||||
Autoimmune connective tissue disease | 3.50 | 1.18 | 10.37 | 0.024 |
| ||||
Trauma/fracture | 2.17 | 1.11 | 4.24 | 0.024 |
| ||||
Immobilization | 3.73 | 2.00 | 6.93 | <.001 |
| ||||
Central vein catheter | 18.73 | 5.36 | 65.48 | <.001 |
| ||||
Any Infections | 3.06 | 2.24 | 4.19 | <.001 |
95% Confidence intervals are calculated using the robust sandwich estimate of Lin and Wei [54] for the covariance and standard error. P-values are Wald p-values from the conditional logistic model using the robust sandwich estimate. Covariates that reached the statistical significance level p≤0.1 are included in this table.
In an unadjusted conditional logistic model we evaluated the association of cancer site and VTE using the cancer site VTE risk score (Table 3a: Model 1).[34] The cancer site VTE risk site score fit most cancers well, but needed some adjustment for lymphoma and MPN (Table 3a: Model 2). Low grade lymphoma (OR=0.34; 95% CI: 0.17, 0.68) was inversely associated with VTE, with lower risk of VTE than would be predicted by the cancer site VTE risk score for lymphoma (see Table 2). In contrast, MPN was positively associated with VTE (OR=18.26; 95% CI: 2.4, 139.1) and had a much higher risk of VTE than predicted by the cancer site VTE risk score (see Table 2; notably, of the 20 patients with MPN, 19 were cases and one was a control). These two cancers were left in the model to improve the fit to the data. After adjusting for these cancer site variables, we then assessed the association between cancer stage and VTE, and noted that the combined variable of stage 2, 3, and 4 versus stage 1 best fit the data (OR=2.38; 95% CI: 1.80, 3.16; Table 3a: Models 3 and 4). After adjusting for the cancer site VTE risk score and the significant cancer site and stage variables, the remaining cancer and non-cancer variables were tested for association with VTE, and the covariates with a p-value ≤ 0.1 are listed in Table 3b.
Table 3a. Univariate Analysis for Association between Venous Thromboembolism in Cancer Patients and Cancer Site, and After Adjusting for Cancer Site VTE Risk Score, Individual Cancer Sites, and Cancer Stage.
Cancer Site and Stage | OR | 95% CI* | p-value* | ||
---|---|---|---|---|---|
Model 1 | Two-fold increase in cancer site VTE risk score | 1.42 | 1.22 | 1.66 | <.001 |
Model 2 | Two-fold increase in cancer site score | 1.52 | 1.28 | 1.82 | <.001 |
Low grade lymphoma | 0.34 | 0.17 | 0.68 | 0.002 | |
Medium or high grade lymphoma | 1.40 | 0.62 | 3.18 | 0.42 | |
Myeloproliferative neoplasm | 18.26 | 2.40 | 139.08 | 0.005 | |
Model 3 | Two-fold increase in cancer site VTE risk score | 1.44 | 1.20 | 1.73 | <.001 |
Low grade lymphoma | 0.30 | 0.15 | 0.61 | <.001 | |
Medium or high grade lymphoma | 1.65 | 0.72 | 3.78 | 0.24 | |
Myeloproliferative neoplasm | 23.96 | 3.05 | 188.00 | 0.002 | |
Stage 3, 4 vs stage 1, 2 | 2.00 | 1.55 | 2.59 | <.001 | |
Model 4 | Two-fold increase in cancer site VTE risk score | 1.50 | 1.25 | 1.81 | <.001 |
Low grade lymphoma | 0.30 | 0.15 | 0.63 | 0.001 | |
Medium or high grade lymphoma | 1.38 | 0.63 | 3.03 | 0.42 | |
Myeloproliferative neoplasm | 28.64 | 3.53 | 232.17 | 0.002 | |
Stage 2, 3, 4 vs stage 1 | 2.38 | 1.80 | 3.16 | <.001 |
95% Confidence intervals are calculated using the robust sandwich estimate of Lin and Wei [54] for the covariance and standard error. P-values are Wald p-values from the conditional logistic model using the robust sandwich estimate. Covariates that reached the statistical significance level p≤0.1 are included in this table.
In the multivariable conditional logistic model, we found that a two-fold increase in the cancer site VTE risk score was independently associated with 1.4-fold increased odds of incident VTE (Table 4a). Cancer stage ≥ 2 independently increased the odds of VTE by 2.2-fold when compared to stage 1. Liver metastasis (OR=2.70), chemotherapy (OR=1.81), progestins (OR=2.13), being underweight (BMI<18.5 kg/m2; OR=1.93) or severely obese (BMI≥35 kg/m2; OR=3.97), hospitalization (OR=7.91), nursing home confinement (OR=4.71), CV catheter use (OR=8.53) and any recent infection (OR=1.71) were independent risk factors for VTE (Table 4a). We did not note any difference in the risk of VTE between the various subclasses of chemotherapy.
Table 4a. Multivariable Analysis for Predictors of Venous Thromboembolism in Cancer Patients.
Characteristic | OR | 95% CI* | P-Value* | |
---|---|---|---|---|
Two-fold increase in Cancer site VTE risk score | 1.37 | 1.09 | 1.71 | 0.007 |
Lymphoma: Low Grade | 0.26 | 0.09 | 0.70 | 0.008 |
Lymphoma: Intermediate or High Grade | 1.42 | 0.50 | 4.03 | 0.51 |
Myeloproliferative neoplasm | 38.99 | 8.72 | 174.26 | <.001 |
Stage 2 or 3 or 4 | 2.18 | 1.56 | 3.02 | <.001 |
Metastasis to liver | 2.70 | 1.69 | 4.31 | <.001 |
Chemotherapy use within 92 days | 1.81 | 1.17 | 2.81 | 0.008 |
Progestin use within 92 days | 2.13 | 1.07 | 4.25 | 0.032 |
Body mass index (BMI) (Reference: 18.5-25 kg/m2) | ||||
BMI <18.5 kg/m2 (Underweight) | 1.93 | 1.02 | 3.66 | 0.044 |
BMI 25-30 kg/m2 (Overweight) | 1.10 | 0.76 | 1.61 | 0.61 |
BMI 30-35 kg/m2 (Obese class I) | 1.67 | 0.98 | 2.86 | 0.062 |
BMI ≥35 kg/m2 (Obese class II-III) | 3.97 | 1.48 | 10.66 | 0.006 |
Location of onset of VTE (Reference: Community) | ||||
Nursing home | 4.71 | 2.01 | 11.04 | <.001 |
Hospitalized within 92 days | 3.89 | 2.53 | 5.97 | <.001 |
Hospitalized | 7.91 | 4.44 | 14.10 | <.001 |
CV catheter use within 92 days | 8.53 | 2.76 | 26.40 | <.001 |
Any infection within 92 days | 1.71 | 1.19 | 2.45 | 0.004 |
95% Confidence intervals are calculated using the robust sandwich estimate of Lin and Wei [54] for the covariance and standard error. P-values are Wald p-values from the conditional logistic model using the robust sandwich estimate.
We explored pre-chemotherapy platelet count ≥350 × 109/L, hemoglobin <100g/L, and leukocyte count >11 × 109/L [40, 43, 44] for an independent association with VTE in the subset of cases and controls where these data were available (Tables 3c and 4b). Of the 1208 cases and controls in this study, data were available on 344 (28.5%) subjects when their case-control pair status was preserved, and in 512 (42.4%) subjects overall. After adjusting for independent risk factors in the final multivariable model, only platelet count ≥350 × 109/L at the time of cancer diagnosis was marginally associated with VTE (OR=2.3; p=0.07; Table 4b).
Table 3c. Univariate Analysis for Predictors of Venous Thromboembolism in Cancer Patients with Indicated Lab Values, Adjusting for Cancer Site VTE Risk Score, Individual Cancer Sites and Cancer Stage (Table 3a).
Characteristic | OR | 95% CI* | p-value* | |
---|---|---|---|---|
| ||||
Leukocyte count >11 × 109/L | ||||
Preserving case-control pair matching† | 1.38 | 0.81 | 2.35 | 0.23 |
| ||||
Not preserving case control pair matching‡ | 1.57 | 0.01 | 172.80 | 0.85 |
| ||||
Hemoglobin <100g/L | ||||
Preserving case-control pair matching† | 0.94 | 0.43 | 2.07 | 0.88 |
| ||||
Not preserving case control pair matching‡ | 1.32 | 0.50 | 3.51 | 0.57 |
| ||||
Platelet count ≥350 × 109/L | ||||
Preserving case-control pair matching† | 1.61 | 0.78 | 3.29 | 0.20 |
| ||||
Not preserving case control pair matching‡ | 1.45 | 0.00 | 1239.26 | 0.91 |
95% Confidence intervals are calculated using the robust sandwich estimate of Lin and Wei [54] for the covariance and standard error. P-values are Wald p-values from the conditional logistic model using the robust sandwich estimate. Covariates that reached the statistical significance level p≤0.1 are included in this table.
Subset of matched cases and controls where data were available in the electronic medical records, preserving the case-control pair status; N=344 sets.
Subset of cases and controls where data were available in the electronic medical records, not preserving the case-control pairing; N=512 individuals.
Table 4b. Multivariable Analysis for Predictors of Venous Thromboembolism in Cancer Patients with Indicated Lab Values.
Characteristic | OR | 95% CI* | P-Value* | |
---|---|---|---|---|
Leukocyte count >11 × 109/L | ||||
Preserving case-control pair matching† | 1.09 | 0.58 | 2.06 | 0.79 |
Not preserving case control pair matching‡ | 1.32 | 0.00 | 517.71 | 0.93 |
Hemoglobin <100g/L | ||||
Preserving case-control pair matching† | 0.51 | 0.17 | 1.53 | 0.23 |
Not preserving case control pair matching‡ | 1.09 | 0.12 | 9.76 | 0.94 |
Platelet count ≥350 × 109/L | ||||
Preserving case-control pair matching† | 2.29 | 0.93 | 5.65 | 0.072 |
Not preserving case control pair matching‡ | 1.54 | 0.00 | 8518.00 | 0.92 |
95% Confidence intervals are calculated using the robust sandwich estimate of Lin and Wei [54] for the covariance and standard error. P-values are Wald p-values from the conditional logistic model using the robust sandwich estimate.
Subset of matched cases and controls where data were available in the electronic medical records, preserving the case-control pair status; N=344 sets.
Subset of cases and controls where data were available in the electronic medical records, not preserving the case-control pairing; N=512 individuals.
To test whether there were changes in VTE events over time due to evolution in clinical practice in cancer diagnosis and therapy during the study period, we tested for and noted no significant interaction between VTE event year and liver metastasis, chemotherapy use, progestin use, VTE location of onset, BMI, CV catheter use, or any infection within 92 days. To test the potential effect of new chemotherapies on VTE risk, we estimated the odds of VTE with chemotherapy use within 92 days over two time frames, 1973-1985 and 1986-2000, and found that the odds were similar in these time frames (OR=1.88 for 1973-1985, and OR=1.74 for 1986-2000), adjusting for all the other covariates in the final multivariable model. Similar odds of VTE were noted for cancer stage ≥ 2 in these two time frames (OR: 2.8 vs. 2.1, respectively). In the time frame 1973-1985, the odds of in-hospital VTE was higher than the odds of VTE within 92 days post-hospitalization (OR: 14.6 vs. 2.8 respectively); in contrast to the latter time frame 1986-2000, where the odds of in-hospital VTE was similar to the odds of VTE within 92 days post-hospitalization (OR: 5.4 vs. 4.6, respectively).
Discussion
We found that cancer site and stage, both measures of biological aggressiveness and procoagulant potential of the underlying cancer, were associated with increased risk of VTE. Our previously-derived cancer site VTE risk score [34], applied as a continuous variable, functioned quite well in predicting the risk of VTE for most cancer sites; a two-fold increase in cancer site VTE risk score increased the risk of VTE nearly 1.4 fold after adjusting for all the other significant covariates listed in Table 4a. While the rank order of SMR for various cancer sites (“cancer site VTE risk score”) among our incident VTE cases [34] was similar to that observed in other studies [6, 7, 15, 31, 32], we uniquely used our score as a continuous variable in this study. This strategy adjusted in a parsimonious way for the magnitude of VTE risk associated with each cancer site which allowed us to focus on other cancer and non-cancer characteristics. However, the cancer site VTE risk score for indolent cancer with long follow-up (e.g., low grade lymphoma and MPN) did not adequately represent the VTE risk, with low grade lymphoma (OR=0.34; Table 3a) being inversely associated with VTE, suggesting that the risk of VTE with low grade lymphoma is lower than would be predicted by the lymphoma cancer site VTE risk score (risk score: 28.8; Table 2). In contrast, MPN was positively associated with VTE (OR=18.26; Table 3a) and had a much higher risk of VTE than predicted by the MPN cancer site VTE risk score (risk score: 12.6; Table 2).
Higher cancer stage has been associated with increased VTE risk.[6, 23, 41] We noted that cancer stage ≥ 2 was independently associated with a 2.2-fold increased odds of VTE. Cancer grade was not associated with VTE risk, as was noted in the Vienna CATS study.[35] Cancer stage progression was not an independent risk factor for VTE after adjustment for cancer stage at cancer diagnosis. Liver metastasis was associated with 2.7-fold increased risk of VTE, independent of cancer stage ≥2 and other cancer and non-cancer risk factors. We speculate that this represents the biological aggressiveness and procoagulant potential of the underlying stage 4 cancer. Another possibility is vascular invasion by the underlying tumor, as patients with primary or metastatic liver cancer are at an increased risk of splanchnic vein thrombosis (n=22; 6.3%).
Recent hospitalization (with or without surgery) and nursing home confinement were associated with nearly 8- and 5-fold increased risks of VTE, respectively. Hospitalization and nursing home confinement likely are surrogate markers of relative immobility and the acuity and severity of their underlying illness. Hospitalization and nursing home confinement have also been identified as independent VTE risk factors in the general population.[2, 31, 37-39] Interestingly, in the time frame 1973-1985, the odds of in-hospital VTE was higher than the odds of VTE within 92 days post-hospitalization (OR: 14.6 vs. 2.8 respectively); in contrast to the latter time frame 1986-2000, where the odds of in-hospital VTE was similar to the odds of VTE within 92 days post-hospitalization (OR: 5.4 vs. 4.6, respectively). This probably reflects the decrease in hospital length of stay over time, and implies that although the VTE events may be occurring at about a similar rate following admission to the hospital, in the recent era, more events are diagnosed following dismissal from the hospital due to shorter length of hospital stay. Although recent surgery was univariately associated with increased risk of VTE in cancer patients, on multivariate analysis, it was no longer an independent risk VTE factor, similar to the Dutch MEGA study [31] but in contrast to the Danish registry.[6] Surgical resection of cancer is typically a treatment modality for lower stage cancers. Furthermore, surgery is offered to individuals who have better performance status and less comorbidity. Interestingly, in a study of patients with colon cancer that utilized California patient discharge data, surgery was associated with a lower risk of VTE.[41]
Recent chemotherapy independently increased the odds of VTE 1.8-fold, as has been noted in other studies.[2, 6, 31, 36, 42] However, we did not detect any difference in VTE risk by type of chemotherapy, possibly due to inadequate power. While cancer therapy continuously evolves over time, the odds of VTE associated with chemotherapy were similar in the time frames, 1973-1985 and 1986-2000 (OR=1.88 and OR=1.75, respectively). As noted in a previous study, radiation therapy was not an independent risk factor for VTE.[31]
Progestin was associated with a two-fold increased risk of VTE, similar to previous studies where progestin was an independent risk factor for VTE in women. [2, 28, 50, 51] We initially postulated that this increased VTE risk would be confined to women where progesterone is typically used for suppression of menstrual bleeding in women undergoing chemotherapy or for treatment of breast or gynecologic cancers. However, the rates of progesterone use in female cases and controls were similar (9.7% and 8.1%, respectively; Table 1). In contrast, in men, 6% of the cases vs. 1% of the controls used progestins, predominantly for treatment of prostate cancer. Although it is biologically plausible that progestins independently increased the risk of VTE in cancer patients, progestin use may indirectly reflect the underlying prostate cancer and the cancer site VTE risk score may be under-adjusting for prostate cancer.
Underweight (BMI<18.5 mg/m2; OR=1.9) or severely obese (BMI≥35 kg/m2; OR=3.97) cancer patients had an increased odds of VTE. Underweight cancer patients may have aggressive cancer, advanced stage disease, cancer cachexia or multiple comorbidities, thus increasing their VTE risk. Moreover, in the general population, being underweight has been associated with poorer survival following incident VTE.[55] On the other hand, severe obesity could predispose to VTE due to the physical effects of body fat impeding the venous return and/or a proinflammatory/ prothrombotic state seen in obesity.[56] Obesity is risk factor for incident VTE in the general population.[57-60]
We were surprised that CV catheter was independently associated with an 8.5-fold increased risk of VTE despite excluding isolated arm and jugular vein DVT. Thus, the risk imparted by CV catheter appears to be more systemic and could reflect chemotherapy and cancer site and/or stage despite adjusting for these characteristics. Recent infection also was independently associated with a 1.7-fold increased VTE risk, possibly due to a procoagulant state induced by infections.[61] Infection is a VTE risk factor for hospitalized cancer patients.[42]
Pre-chemotherapy platelet count ≥350 × 109/L, hemoglobin <100g/L, and leukocyte count >11 × 109/L are reported as risk factors for cancer-associated VTE within three months after starting chemotherapy.[40, 43, 44] In a subset of our cases and controls where complete blood count data at the time of cancer diagnosis were available, only platelet count ≥350 × 109/L was marginally associated with VTE in the final multivariable model, likely due to differences in study design. In the Awareness of Neutropenia in Cancer Study Group Registry of 4066 cancer patients who were initiating a new chemotherapy regimen, there were a total of 88 VTE cases during a median follow-up of 2.4 months; the distribution of cancer types did not reflect the distribution of all the cancers occurring in the community (i.e., two-thirds had either breast, lung or colon cancer). [40]_The Vienna CATS cohort study included 819 either newly diagnosed cancer patients or those with progression of cancer who had not recently received therapy, of whom 61 developed VTE over a median follow-up of 656 days. [44]_Our study was a matched case-control study of subjects with active cancer, irrespective of whether they were receiving therapy, as one of the goals was to analyze whether chemotherapy use and type, radiation therapy and/ or surgical resection are independent VTE risk factors in cancer patients. As the complete blood count data were available on 172 VTE case-control pairs (i.e., 344 subjects) and 512 of 1208 subjects overall, we do not think that the differences are due to limited numbers in the subset analysis.
Our study has several important strengths. Due to the unique features of the REP, our study avoids referral bias. Our population-based study better reflects “real world” clinical practice in contrast to a highly selected population participating in clinical trials. All VTE cases met strict criteria for clinically or objectively-confirmed acute DVT and/or PE, and we confirmed that controls did not have VTE, based on direct review of their source documents (i.e., imaging, surgical and autopsy reports) rather than depending on administrative codes. We included only individuals who had active cancer, and the definition of active cancer was explicit. We included the entire spectrum of VTE disease occurring in the community, including persons with rapidly fatal and chronic care facility (e.g., nursing home) events who did not reach the hospital. We ensured a comparable control group by performing a population-based study where both cases and controls were residents from the same community with similar duration of active cancer, thus avoiding the issues of ‘immortal time bias’ [62], and lifetime access to medical care. By using the cancer site SMR as a continuous variable “cancer site VTE risk score” we were able to jointly adjust for cancer site and stage and quantify the magnitude of VTE risk associated with each of the risk factors identified after adjusting for the other risk factors. Although there has been evolution in clinical practice in cancer diagnosis and therapy during the study period, we did not note a significant interaction between VTE event year and the identified VTE risk factors including chemotherapy use, VTE location of onset, BMI, CV catheter use, or any infection.
Our study also has important limitations. The age-, sex- and racial-distribution of Olmsted County is similar to that for Minnesota, the upper mid-west, and the U.S. white population; however, residents of Olmsted County exhibit higher median income and education level compared to these geographic regions.[46, 63, 64] While no single geographic area is representative of all others, the under-representation of minorities may compromise the generalizability of our findings to different racial and ethnic groups. Due to matching, patient age and sex could not be tested as a main effect for an association with cancer-associated incident VTE although we did test these for potential interactions and found none. Patients who had common cancers or who survived longer may have been more likely to be chosen as controls. Due to the generally lower autopsy rate in older individuals and the declining autopsy rate over the study period (the autopsy rates for men and women for 1973-1980, 1981-1990 and 1991-2000 were around 50%, 40%, 35% and 35%, 30% and 25%, respectively) [65], we may have missed fatal PE cases presenting as sudden death, as suggested by the declining number of PE’s that were discovered on autopsy over time. Finally, due to our study timeframe we could not confirm the association of immune modulating drugs (e.g., thalidomide, lenalidomide) or angiogenesis inhibitors (e.g., bevacizumab) with VTE. Adjusting for cancer and non-cancer characteristics that we determined to be associated with VTE independent of chemotherapy will be important for future studies testing new chemotherapies for an association with VTE. However, despite the introduction of newer agents and potential changes in the standard chemotherapy regimens over time, the findings of this study are still currently relevant and applicable today as we have identified and quantified clinically important cancer and non-cancer related risk factors that are independently associated with VTE.
In conclusion, cancer site, as represented by the cancer site VTE risk score, cancer stage ≥2, liver metastasis, chemotherapy and progestin use, being underweight or obese, central venous catheter, hospitalization, nursing home confinement and infection are independent risk factors for incident VTE in individuals with active cancer. These characteristics can identify high VTE-risk active cancer patients for development of a risk score and for inclusion in future studies of VTE prophylaxis.
Key Points.
We tested cancer and non-cancer characteristics as predictors of cancer-associated VTE
Cancer site, stage, liver metastasis, chemotherapy, progestin increased VTE risk
BMI, hospitalization, central venous catheter and infection increased VTE risk
These characteristics can be used to identify high VTE-risk active cancer patients
Acknowledgements
We gratefully acknowledge Catherine L. Brandel, Diadra H. Else, Jane A. Emerson, and Cynthia L. Nosek for excellent data collection. Research reported in this publication was supported by a grant from the National Institutes of Health, National Heart Lung and Blood Institute (R01HL66216 to J.A.H.) and was made possible by the Rochester Epidemiology Project (National Institutes of Health, National Institute on Aging (R01AG034676). Research support also was provided by Mayo Foundation. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Statement of Prior Presentation: Presented in abstract form at the XXV Congress of International Society on Thrombosis and Haemostasis, Toronto, Canada, June 23, 2015 (OR128).
Authorship Contributions:
A.A.A. designed and performed the research, collected, analyzed, and interpreted the data and wrote the manuscript; R.E.G. collected data, performed the statistical analyses, and contributed to the manuscript; T.M.P. designed and performed the research, collected data, performed the statistical analyses, and contributed to the manuscript; R.S.M. collected and interpreted the data, and contributed to the manuscript; K.R.B. contributed to the study design, directed the statistical analyses, and contributed to the manuscript; and J.A.H. designed and performed the research, collected, analyzed and interpreted the data, and wrote the manuscript.
Disclosure of Conflicts of Interest:
The authors have no competing financial or conflict of interest to declare.
References
- 1.Cogo A, Bernardi E, Prandoni P, Girolami B, Noventa F, Simioni P, Girolami A. Acquired risk factors for deep-vein thrombosis in symptomatic outpatients. Arch Intern Med. 1994;154(2):164–8. [PubMed] [Google Scholar]
- 2.Heit JA, Silverstein MD, Mohr DN, Petterson TM, O’Fallon WM, Melton LJ., 3rd Risk factors for deep vein thrombosis and pulmonary embolism: a population-based case-control study. Arch Intern Med. 2000;160(6):809–15. doi: 10.1001/archinte.160.6.809. [DOI] [PubMed] [Google Scholar]
- 3.Rosendaal FR. Risk factors for venous thrombotic disease. Thromb Haemost. 1999;82(2):610–9. [PubMed] [Google Scholar]
- 4.Samama MM. An epidemiologic study of risk factors for deep vein thrombosis in medical outpatients: the Sirius study. Arch Intern Med. 2000;160(22):3415–20. doi: 10.1001/archinte.160.22.3415. [DOI] [PubMed] [Google Scholar]
- 5.Blom JW, Doggen CJ, Osanto S, Rosendaal FR. Malignancies, prothrombotic mutations, and the risk of venous thrombosis. Jama. 2005;293(6):715–22. doi: 10.1001/jama.293.6.715. [DOI] [PubMed] [Google Scholar]
- 6.Cronin-Fenton DP, Sondergaard F, Pedersen LA, Fryzek JP, Cetin K, Acquavella J, Baron JA, Sorensen HT. Hospitalisation for venous thromboembolism in cancer patients and the general population: a population-based cohort study in Denmark, 1997-2006. Br J Cancer. 2010;103(7):947–53. doi: 10.1038/sj.bjc.6605883. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Walker AJ, Card TR, West J, Crooks C, Grainge MJ. Incidence of venous thromboembolism in patients with cancer - a cohort study using linked United Kingdom databases. European journal of cancer. 2013;49(6):1404–13. doi: 10.1016/j.ejca.2012.10.021. [DOI] [PubMed] [Google Scholar]
- 8.Heit JA, O’Fallon WM, Petterson TM, Lohse CM, Silverstein MD, Mohr DN, Melton LJ., 3rd Relative impact of risk factors for deep vein thrombosis and pulmonary embolism: a population-based study. Arch Intern Med. 2002;162(11):1245–8. doi: 10.1001/archinte.162.11.1245. [DOI] [PubMed] [Google Scholar]
- 9.Braekkan SK, Borch KH, Mathiesen EB, Njolstad I, Wilsgaard T, Hansen JB. Body height and risk of venous thromboembolism: The Tromso Study. American journal of epidemiology. 2010;171(10):1109–15. doi: 10.1093/aje/kwq066. [DOI] [PubMed] [Google Scholar]
- 10.Spencer FA, Lessard D, Emery C, Reed G, Goldberg RJ. Venous thromboembolism in the outpatient setting. Archives of internal medicine. 2007;167(14):1471–5. doi: 10.1001/archinte.167.14.1471. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.White RH, Zhou H, Murin S, Harvey D. Effect of ethnicity and gender on the incidence of venous thromboembolism in a diverse population in California in 1996. Thromb Haemost. 2005;93(2):298–305. doi: 10.1160/TH04-08-0506. [DOI] [PubMed] [Google Scholar]
- 12.Donati MB. Cancer and thrombosis. Haemostasis. 1994;24(2):128–31. doi: 10.1159/000217092. [DOI] [PubMed] [Google Scholar]
- 13.Arkel YS. Thrombosis and cancer. Seminars in oncology. 2000;27(3):362–74. [PubMed] [Google Scholar]
- 14.Khorana AA, Francis CW, Culakova E, Kuderer NM, Lyman GH. Thromboembolism is a leading cause of death in cancer patients receiving outpatient chemotherapy. J Thromb Haemost. 2007;5(3):632–4. doi: 10.1111/j.1538-7836.2007.02374.x. [DOI] [PubMed] [Google Scholar]
- 15.Levitan N, Dowlati A, Remick SC, Tahsildar HI, Sivinski LD, Beyth R, Rimm AA. Rates of initial and recurrent thromboembolic disease among patients with malignancy versus those without malignancy. Risk analysis using Medicare claims data. Medicine (Baltimore) 1999;78(5):285–91. doi: 10.1097/00005792-199909000-00001. [DOI] [PubMed] [Google Scholar]
- 16.Hutten BA, Prins MH, Gent M, Ginsberg J, Tijssen JG, Buller HR. Incidence of recurrent thromboembolic and bleeding complications among patients with venous thromboembolism in relation to both malignancy and achieved international normalized ratio: a retrospective analysis. J Clin Oncol. 2000;18(17):3078–83. doi: 10.1200/JCO.2000.18.17.3078. [DOI] [PubMed] [Google Scholar]
- 17.Palareti G, Legnani C, Lee A, Manotti C, Hirsh J, D’Angelo A, Pengo V, Moia M, Coccheri S. A comparison of the safety and efficacy of oral anticoagulation for the treatment of venous thromboembolic disease in patients with or without malignancy. Thromb Haemost. 2000;84(5):805–10. [PubMed] [Google Scholar]
- 18.Prandoni P, Lensing AW, Piccioli A, Bernardi E, Simioni P, Girolami B, Marchiori A, Sabbion P, Prins MH, Noventa F, Girolami A. Recurrent venous thromboembolism and bleeding complications during anticoagulant treatment in patients with cancer and venous thrombosis. Blood. 2002;100(10):3484–8. doi: 10.1182/blood-2002-01-0108. [DOI] [PubMed] [Google Scholar]
- 19.Elting LS, Escalante CP, Cooksley C, Avritscher EB, Kurtin D, Hamblin L, Khosla SG, Rivera E. Outcomes and cost of deep venous thrombosis among patients with cancer. Arch Intern Med. 2004;164(15):1653–61. doi: 10.1001/archinte.164.15.1653. [DOI] [PubMed] [Google Scholar]
- 20.Gross CP, Galusha DH, Krumholz HM. The impact of venous thromboembolism on risk of death or hemorrhage in older cancer patients. Journal of general internal medicine. 2007;22(3):321–6. doi: 10.1007/s11606-006-0019-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Ruiz-Gimenez N, Suarez C, Gonzalez R, Nieto JA, Todoli JA, Samperiz AL, Monreal M. Predictive variables for major bleeding events in patients presenting with documented acute venous thromboembolism. Findings from the RIETE Registry. Thrombosis and haemostasis. 2008;100(1):26–31. doi: 10.1160/TH08-03-0193. [DOI] [PubMed] [Google Scholar]
- 22.Sorensen HT, Mellemkjaer L, Olsen JH, Baron JA. Prognosis of cancers associated with venous thromboembolism. N Engl J Med. 2000;343(25):1846–50. doi: 10.1056/NEJM200012213432504. [DOI] [PubMed] [Google Scholar]
- 23.Chew HK, Wun T, Harvey D, Zhou H, White RH. Incidence of venous thromboembolism and its effect on survival among patients with common cancers. Arch Intern Med. 2006;166(4):458–64. doi: 10.1001/archinte.166.4.458. [DOI] [PubMed] [Google Scholar]
- 24.Monreal M, Falga C, Valdes M, Suarez C, Gabriel F, Tolosa C, Montes J. Fatal pulmonary embolism and fatal bleeding in cancer patients with venous thromboembolism: findings from the RIETE registry. Journal of thrombosis and haemostasis: JTH. 2006;4(9):1950–6. doi: 10.1111/j.1538-7836.2006.02082.x. [DOI] [PubMed] [Google Scholar]
- 25.Naess IA, Christiansen SC, Romundstad P, Cannegieter SC, Rosendaal FR, Hammerstrom J. Incidence and mortality of venous thrombosis: a population-based study. J Thromb Haemost. 2007;5(4):692–9. doi: 10.1111/j.1538-7836.2007.02450.x. [DOI] [PubMed] [Google Scholar]
- 26.Dentali F, Ageno W, Pierfranceschi MG, Imberti D, Malato A, Nitti C, Salvi A, Siragusa S, Squizzato A, Vitale J, Agnelli G. Prognostic relevance of an asymptomatic venous thromboembolism in patients with cancer. Journal of thrombosis and haemostasis: JTH. 2011;9(5):1081–3. doi: 10.1111/j.1538-7836.2011.04259.x. [DOI] [PubMed] [Google Scholar]
- 27.Gussoni G, Frasson S, La Regina M, Di Micco P, Monreal M. Three-month mortality rate and clinical predictors in patients with venous thromboembolism and cancer. Findings from the RIETE registry. Thromb Res. 2013;131(1):24–30. doi: 10.1016/j.thromres.2012.10.007. [DOI] [PubMed] [Google Scholar]
- 28.Chee CE, Ashrani AA, Marks RS, Petterson TM, Bailey KR, Melton LJ, 3rd, Heit JA. Predictors of venous thromboembolism recurrence and bleeding among active cancer patients: a population-based cohort study. Blood. 2014;123(25):3972–8. doi: 10.1182/blood-2014-01-549733. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Thodiyil PA, Kakkar AK. Variation in relative risk of venous thromboembolism in different cancers. Thromb Haemost. 2002;87(6):1076–7. [PubMed] [Google Scholar]
- 30.Sallah S, Wan JY, Nguyen NP. Venous thrombosis in patients with solid tumors: determination of frequency and characteristics. Thromb Haemost. 2002;87(4):575–9. [PubMed] [Google Scholar]
- 31.Blom JW, Vanderschoot JP, Oostindier MJ, Osanto S, van der Meer FJ, Rosendaal FR. Incidence of venous thrombosis in a large cohort of 66,329 cancer patients: results of a record linkage study. J Thromb Haemost. 2006;4(3):529–35. doi: 10.1111/j.1538-7836.2006.01804.x. [DOI] [PubMed] [Google Scholar]
- 32.Horsted F, West J, Grainge MJ. Risk of venous thromboembolism in patients with cancer: a systematic review and meta-analysis. PLoS medicine. 2012;9(7):e1001275. doi: 10.1371/journal.pmed.1001275. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Timp JF, Braekkan SK, Versteeg HH, Cannegieter SC. Epidemiology of cancer-associated venous thrombosis. Blood. 2013;122(10):1712–23. doi: 10.1182/blood-2013-04-460121. [DOI] [PubMed] [Google Scholar]
- 34.Petterson TM, Marks RS, Ashrani AA, Bailey KR, Heit JA. Risk of site-specific cancer in incident venous thromboembolism: A population-based study. Thromb Res. 2014;135(3):472–478. doi: 10.1016/j.thromres.2014.12.013. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Ahlbrecht J, Dickmann B, Ay C, Dunkler D, Thaler J, Schmidinger M, Quehenberger P, Haitel A, Zielinski C, Pabinger I. Tumor grade is associated with venous thromboembolism in patients with cancer: results from the Vienna Cancer and Thrombosis Study. Journal of clinical oncology: official journal of the American Society of Clinical Oncology. 2012;30(31):3870–5. doi: 10.1200/JCO.2011.40.1810. [DOI] [PubMed] [Google Scholar]
- 36.Otten HM, Mathijssen J, ten Cate H, Soesan M, Inghels M, Richel DJ, Prins MH. Symptomatic venous thromboembolism in cancer patients treated with chemotherapy: an underestimated phenomenon. Arch Intern Med. 2004;164(2):190–4. doi: 10.1001/archinte.164.2.190. [DOI] [PubMed] [Google Scholar]
- 37.Nalluri SR, Chu D, Keresztes R, Zhu X, Wu S. Risk of venous thromboembolism with the angiogenesis inhibitor bevacizumab in cancer patients: a meta-analysis. Jama. 2008;300(19):2277–85. doi: 10.1001/jama.2008.656. [DOI] [PubMed] [Google Scholar]
- 38.Bennett CL, Silver SM, Djulbegovic B, Samaras AT, Blau CA, Gleason KJ, Barnato SE, Elverman KM, Courtney DM, McKoy JM, Edwards BJ, Tigue CC, Raisch DW, Yarnold PR, Dorr DA, Kuzel TM, Tallman MS, Trifilio SM, West DP, Lai SY, Henke M. Venous thromboembolism and mortality associated with recombinant erythropoietin and darbepoetin administration for the treatment of cancer-associated anemia. Jama. 2008;299(8):914–24. doi: 10.1001/jama.299.8.914. [DOI] [PubMed] [Google Scholar]
- 39.Ashrani AA, Rajkumar SV. Chemotherapy-associated thrombosis. Cancer treatment and research. 2009;148:181–206. doi: 10.1007/978-0-387-79962-9_11. [DOI] [PubMed] [Google Scholar]
- 40.Khorana AA, Francis CW, Culakova E, Lyman GH. Risk factors for chemotherapy-associated venous thromboembolism in a prospective observational study. Cancer. 2005;104(12):2822–9. doi: 10.1002/cncr.21496. [DOI] [PubMed] [Google Scholar]
- 41.Alcalay A, Wun T, Khatri V, Chew HK, Harvey D, Zhou H, White RH. Venous thromboembolism in patients with colorectal cancer: incidence and effect on survival. J Clin Oncol. 2006;24(7):1112–8. doi: 10.1200/JCO.2005.04.2150. [DOI] [PubMed] [Google Scholar]
- 42.Khorana AA, Francis CW, Culakova E, Kuderer NM, Lyman GH. Frequency, risk factors, and trends for venous thromboembolism among hospitalized cancer patients. Cancer. 2007;110(10):2339–46. doi: 10.1002/cncr.23062. [DOI] [PubMed] [Google Scholar]
- 43.Khorana AA, Kuderer NM, Culakova E, Lyman GH, Francis CW. Development and validation of a predictive model for chemotherapy-associated thrombosis. Blood. 2008;111(10):4902–7. doi: 10.1182/blood-2007-10-116327. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44.Ay C, Dunkler D, Marosi C, Chiriac AL, Vormittag R, Simanek R, Quehenberger P, Zielinski C, Pabinger I. Prediction of venous thromboembolism in cancer patients. Blood. 2010;116(24):5377–82. doi: 10.1182/blood-2010-02-270116. [DOI] [PubMed] [Google Scholar]
- 45.Mandala M, Clerici M, Corradino I, Vitalini C, Colombini S, Torri V, De Pascale A, Marsoni S. Incidence, risk factors and clinical implications of venous thromboembolism in cancer patients treated within the context of phase I studies: the ‘SENDO experience’. Annals of oncology: official journal of the European Society for Medical Oncology / ESMO. 2012;23(6):1416–21. doi: 10.1093/annonc/mdr524. [DOI] [PubMed] [Google Scholar]
- 46.Melton LJ., 3rd History of the Rochester Epidemiology Project. Mayo Clin Proc. 1996;71(3):266–74. doi: 10.4065/71.3.266. [DOI] [PubMed] [Google Scholar]
- 47.Silverstein MD, Heit JA, Mohr DN, Petterson TM, O’Fallon WM, Melton LJ., 3rd Trends in the incidence of deep vein thrombosis and pulmonary embolism: a 25-year population-based study. Arch Intern Med. 1998;158(6):585–93. doi: 10.1001/archinte.158.6.585. [DOI] [PubMed] [Google Scholar]
- 48.Heit JA, Sobell JL, Li H, Sommer SS. The incidence of venous thromboembolism among Factor V Leiden carriers: a community-based cohort study. J Thromb Haemost. 2005;3(2):305–11. doi: 10.1111/j.1538-7836.2004.01117.x. [DOI] [PubMed] [Google Scholar]
- 49.Heit JA, Lahr BD, Ashrani AA, Petterson TM, Bailey KR. Predictors of Venous Thromboembolism Recurrence, Adjusted for Treatments and Interim Exposures: A Population-based Case-cohort Study. Thromb Res. 2015;136(2):298–307. doi: 10.1016/j.thromres.2015.06.030. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 50.Heit JA, Leibson CL, Ashrani AA, Petterson TM, Bailey KR, Melton LJ., 3rd Is diabetes mellitus an independent risk factor for venous thromboembolism?: a population-based case-control study. Arteriosclerosis, thrombosis, and vascular biology. 2009;29(9):1399–405. doi: 10.1161/ATVBAHA.109.189290. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 51.Barsoum MK, Heit JA, Ashrani AA, Leibson CL, Petterson TM, Bailey KR. Is progestin an independent risk factor for incident venous thromboembolism? A population-based case-control study. Thrombosis research. 2010;126(5):373–8. doi: 10.1016/j.thromres.2010.08.010. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 52.Ashrani AA, Barsoum MK, Crusan DJ, Petterson TM, Bailey KR, Heit JA. Is lipid lowering therapy an independent risk factor for venous thromboembolism? A population-based case-control study. Thrombosis Research. 2015;136(6):1110–1116. doi: 10.1016/j.thromres.2015.04.005. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 53.Greene F.l., Page DL, Fleming ID, Fritz AG, Balch CM, Haller DG, Morrow M., editors. AJCC Cancer Staging Manual. Sixth Springer Science; 2002. [Google Scholar]
- 54.Lin DY, Wei LJ. The Robust Inference for the Proportional Hazards Model. Journal of the American Statistical Association. 1989;84:1074–1078. [Google Scholar]
- 55.Heit JA, Silverstein MD, Mohr DN, Petterson TM, O’Fallon WM, Melton LJ., 3rd Predictors of survival after deep vein thrombosis and pulmonary embolism: a population-based, cohort study. Arch Intern Med. 1999;159(5):445–53. doi: 10.1001/archinte.159.5.445. [DOI] [PubMed] [Google Scholar]
- 56.Allman-Farinelli MA. Obesity and venous thrombosis: a review. Seminars in thrombosis and hemostasis. 2011;37(8):903–7. doi: 10.1055/s-0031-1297369. [DOI] [PubMed] [Google Scholar]
- 57.Goldhaber SZ, Savage DD, Garrison RJ, Castelli WP, Kannel WB, McNamara PM, Gherardi G, Feinleib M. Risk factors for pulmonary embolism. The Framingham Study. The American journal of medicine. 1983;74(6):1023–8. doi: 10.1016/0002-9343(83)90805-7. [DOI] [PubMed] [Google Scholar]
- 58.Hansson PO, Eriksson H, Welin L, Svardsudd K, Wilhelmsen L. Smoking and abdominal obesity: risk factors for venous thromboembolism among middle-aged men: “the study of men born in 1913”. Arch Intern Med. 1999;159(16):1886–90. doi: 10.1001/archinte.159.16.1886. [DOI] [PubMed] [Google Scholar]
- 59.Abdollahi M, Cushman M, Rosendaal FR. Obesity: risk of venous thrombosis and the interaction with coagulation factor levels and oral contraceptive use. Thromb Haemost. 2003;89(3):493–8. [PubMed] [Google Scholar]
- 60.Stein PD, Beemath A, Olson RE. Obesity as a risk factor in venous thromboembolism. The American journal of medicine. 2005;118(9):978–80. doi: 10.1016/j.amjmed.2005.03.012. [DOI] [PubMed] [Google Scholar]
- 61.Levi M, van der Poll T, Buller HR. Bidirectional relation between inflammation and coagulation. Circulation. 2004;109(22):2698–704. doi: 10.1161/01.CIR.0000131660.51520.9A. [DOI] [PubMed] [Google Scholar]
- 62.Suissa S. Immortal time bias in pharmaco-epidemiology. American journal of epidemiology. 2008;167(4):492–9. doi: 10.1093/aje/kwm324. [DOI] [PubMed] [Google Scholar]
- 63.St Sauver JL, Grossardt BR, Yawn BP, Melton LJ, 3rd, Rocca WA. Use of a medical records linkage system to enumerate a dynamic population over time: the Rochester epidemiology project. American journal of epidemiology. 2011;173(9):1059–68. doi: 10.1093/aje/kwq482. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 64.St Sauver JL, Grossardt BR, Leibson CL, Yawn BP, Melton LJ, 3rd, Rocca WA. Generalizability of epidemiological findings and public health decisions: an illustration from the Rochester Epidemiology Project. Mayo Clinic Proceedings. 2012;87(2):151–60. doi: 10.1016/j.mayocp.2011.11.009. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 65.St Sauver JL, Grossardt BR, Yawn BP, Melton LJ, 3rd, Pankratz JJ, Brue SM, et al. Data resource profile: the Rochester Epidemiology Project (REP) medical records-linkage system. Int J Epidemiol. 2012 Dec;41(6):1614–24. doi: 10.1093/ije/dys195. [DOI] [PMC free article] [PubMed] [Google Scholar]