Fax: (585) 273-1042
Risk factors for chemotherapy-associated venous thromboembolism in a prospective observational study
Version of Record online: 13 NOV 2005
Copyright © 2005 American Cancer Society
Volume 104, Issue 12, pages 2822–2829, 15 December 2005
How to Cite
Khorana, A. A., Francis, C. W., Culakova, E. and Lyman, G. H. (2005), Risk factors for chemotherapy-associated venous thromboembolism in a prospective observational study. Cancer, 104: 2822–2829. doi: 10.1002/cncr.21496
- Issue online: 8 DEC 2005
- Version of Record online: 13 NOV 2005
- James P. Wilmot Cancer Research Fellowship
- The Awareness of Neutropenia in Cancer (ANC) Study Group Registry is funded by Amgen, Inc.
- venous thromboembolism;
The incidence of venous thromboembolism (VTE) is increased in cancer, but little information is available about risk factors in cancer patients on chemotherapy.
We analyzed data from a prospective, multicenter observational study to determine the frequency and risk factors for VTE in ambulatory cancer patients initiating a new chemotherapy regimen. The association of VTE with clinical variables was characterized using univariate and multivariate analysis.
Among 3003 patients treated with at least one cycle of chemotherapy, VTE occurred in 58 (1.93%) over a median follow-up of 2.4 months (0.8%/mo). The incidence varied significantly by site of cancer (P = 0.01) with highest rates in upper gastrointestinal (2.3%/mo) and lung cancer (1.2%/mo), and lymphoma (1.1%/mo). An elevated prechemotherapy platelet count was significantly associated with an increased rate of VTE (P for trend = 0.005). The incidence of VTE was 3.98% (1.66%/mo) for patients with a prechemotherapy platelet count ≥ 350,000, compared with 1.25% (0.52%/mo) for patients with platelet counts of < 200,000 (P for trend=0.0003). In multivariate analysis, a prechemotherapy platelet count of ≥ 350,000/mm3 (adjusted OR 2.81, 95% CI 1.63–4.93, P = 0.0002), site of cancer, hemoglobin < 10g/dL or use of erythropoietin, and use of white cell growth factors in high-risk sites of cancer were significantly associated with VTE.
Symptomatic VTE is a frequent complication of chemotherapy. The prechemotherapy platelet count is a unique risk factor and can help identify high-risk patients for future trials of thromboprophylaxis. Cancer 2005. © 2005 American Cancer Society.
The association between thrombosis and malignancy has been known for centuries, with the earliest reference dating to the Indian surgeon Sushruta who lived in approximately 1000 BC.1 Armand Trousseau of France was the first to comprehensively describe the eponymous syndrome in 1865 and to suggest that thrombosis in malignancy was more than an epiphenomenon.1, 2 It is now well recognized that the risk of VTE is increased in cancer patients, especially in those receiving chemotherapy.3–6 Newer anticancer therapies, in particular antiangiogenic agents, are associated with very high rates of VTE.7–9 The estimated annual incidence of VTE in the cancer population is 0.5% (0.04%/mo), compared with 0.1% in the general population.10 The diagnosis of VTE in cancer has important clinical implications, as it is the second leading cause of death in cancer patients.11–13 Indeed, evidence suggests that elements of the coagulation cascade including tissue factor, thrombin, and fibrinogen can enhance tumor growth, metastasis, and angiogenesis.14–17 Cancer diagnosed at the same time as or within 1 year of an episode of VTE is associated with an advanced stage and a threefold lower survival at 1 year.18 We have recently shown that hospitalized cancer patients experience a greater in-hospital mortality rate if they develop VTE (odds ratio 2.01, 95% CI 1.83–2.22, P < 0.0001), even without metastatic disease.19 In addition, the occurrence of VTE may have consequences related to patient morbidity, interruption of chemotherapy, and cost of hospitalization.20
Contemporary studies characterizing the incidence of VTE in cancer patients are scarce, with most available literature comprising either retrospective studies or post-hoc analyses of clinical trials. A study using Medicare claims data for hospital discharges from 1988–90 indicated that cancer patients developed VTE during hospitalization significantly more frequently than noncancer patients, although incidence rates were only 0.6% and 0.57%, respectively.21 A more recent study found that 7.8% of cancer patients treated at three medical centers developed VTE over a median follow-up of 26 months.22 Incidence rates of VTE during adjuvant therapy of breast cancer vary widely from 2.1% to 13.6%.6, 23 Major advances have been made in the past several years in understanding risk factors for VTE in the general population, and some can be extrapolated to cancer-associated VTE.4 In cancer patients, additional risk factors may include the stage and site of disease, use of chemotherapy, and the presence of a central venous catheter.24, 25 These associations are based on retrospective analyses. A prospective study of VTE in cancer patients receiving chemotherapy has not been conducted, and there is limited information concerning risk factors that specifically predispose to VTE in this patient group.24 In the only clinical trial of VTE prophylaxis during chemotherapy, patients with metastatic breast cancer who received very-low-dose warfarin demonstrated a reduced incidence of VTE compared with placebo.26 Current guidelines do not recommend VTE prophylaxis for ambulatory cancer patients,24 because the majority of cancer patients do not develop VTE, and because the use of anticoagulants in cancer patients is associated with an increased risk of bleeding complications.27 Indeed, evidence indicates that medical oncologists rarely provide thromboprophylaxis to cancer patients.28 The ability to stratify risk would allow appropriate use of VTE prophylaxis only in patients at highest risk.
We, therefore, chose to determine the occurrence and risk factors for symptomatic VTE using data from an ongoing prospective observational study of cancer patients initiating a new chemotherapy regimen. This study population was relatively homogenous, because cancer patients with acute medical illnesses, hospitalization, or terminal conditions who are already at high risk for development of VTE were not included. This allowed us to study risk factors associated primarily with the patient's cancer and cancer-related therapy. We further determined the effects of multiple clinical and laboratory covariates on VTE incidence to identify a population of ambulatory patients at high risk for VTE.
MATERIALS AND METHODS
Study Design and Population
The study population comprised consecutively enrolled patients in the Awareness of Neutropenia in Cancer (ANC) Study Group Registry, an observational multicenter study designed to evaluate febrile neutropenia and other chemotherapy-related complications in cancer patients who are beginning a new chemotherapy regimen. Patients were followed prospectively for a maximum of four cycles. Patients enrolled between March 2002 and August 2004 who had completed at least one cycle of chemotherapy were included in this analysis. Patients were enrolled at 115 sites within the United States, balanced for practice volume and geographic location. The study was approved by a central institutional review board (IRB) as well as the University of Rochester IRB, and patients provided informed consent.
Patients were required to have a histologically confirmed diagnosis of cancer, with targeted enrollment of specific tumor types (breast, lung, ovarian, sarcoma, colon, and lymphomas), to be at the start of a new chemotherapy regimen, be of age 18 years or older, and capable of providing informed consent. Patients were excluded if they were receiving concurrent cytotoxic, biologic, or immunologic therapy for other conditions, or continuous single agent chemotherapy, if they had a diagnosis of acute leukemia or myeloma, were pregnant or lactating, had an active infection requiring treatment, were currently participating in a double-blind study, or had received stem cell transplantation.
Data was collected prospectively for up to four cycles of chemotherapy. This included baseline information including current medications, recent surgery, comorbidities, a complete blood count, the planned chemotherapy regimen for the patient, including the individual drugs and doses being used, the dosing interval within the cycle, dosing interval between cycles (length of cycle), and the total number of cycles planned. Data, including serial complete blood counts, changes made to the planned chemotherapy treatment, including dose reduction or discontinuation for individual drugs within the regimen, as well as a change in the chemotherapy regimen itself, were collected during visits at the beginning and nadir of each new cycle and during unexpected midcycle visits. VTE was diagnosed by the treating clinician on the basis of clinical suspicion by using usual diagnostic procedures. The occurrence of a symptomatic VTE event after entering the study was reported in the new-cycle visit forms, or in the midcycle visit forms, depending upon the timing of the event.
The chi-square test was used to compare categorical variables, and the Cochran–Armitage test was used to determine trend across multiple ordered categories. The association of thromboembolism with clinical variables was reported as a rate and evaluated in univariate analysis using a chi-square test. Variables associated with an increased risk of VTE (P ≤ 0.05) in univariate analysis were included in a multivariate logistic regression model. First-order interaction terms for the primary outcome of VTE were explored and reported.
The study population comprised 3003 patients enrolled between March 2002 and August 2004. The average age was 60 years and over one-third were older than 65 years (Table 1). Breast cancer was the most common diagnosis, followed by lung and colon cancer. Non-Hodgkin lymphoma was the most common hematologic malignancy. Women were predominant in the population because of the high enrollment of breast cancer patients. Over one-third of patients had metastatic disease at the time of study entry. Nearly one-fourth of patients had one or more comorbidities.
|All patients||3003 (100)|
|< 65 yrs||1840 (61.5)|
|≥ 65 yrs||1152 (38.5)|
|ECOG performance status|
|Site of cancer|
|Upper gastrointestinal||89 (3)|
|Non-Hodgkin lymphoma||267 (8.9)|
|Hodgkin disease||49 (1.6)|
|Diabetes mellitus||356 (11.9)|
|Pulmonary disease||235 (7.9)|
|Myocardial infarction||102 (3.4)|
|Peripheral vascular disease||71 (2.4)|
|Congestive heart failure||69 (2.3)|
|Cerebrovascular disease||51 (1.7)|
|Renal disease||33 (1.1)|
|Body surface area ≥ 2 m2||803 (26.7)|
|Baseline hematologic parameters|
|Platelets ≥ 350,000/mm3||653 (21.9)|
|Platelets 200-350,000/mm3||1,769 (59.3)|
|Platelets < 200,000/mm3||559 (18.8)|
|Hemoglobin < 10 g/dL||179 (6)|
|Use of growth factors during Cycle 1|
|White blood cell growth factors||1007 (33.5)|
|Red blood cell growth factors||791 (26.3)|
|Type of regimen|
|Other cancer treatment|
|Hormonal therapy||212 (7.1)|
A total of 58 patients (1.93%) developed VTE during a median follow-up period of 2.4 months (0.8%/mo). Deep venous thrombosis (DVT) developed in 47 patients, pulmonary embolism (PE) in 14, and concurrent DVT and PE in 3 patients. One patient developed VTE before starting chemotherapy. The rate of VTE did not differ significantly among chemotherapy cycles, occurring in 0.77% during Cycle 1, 0.74% during Cycle 2, and 0.7% during Cycle 3. The cumulative rate of VTE was 2.22% (95% CI, 1.65–2.79) during Cycles 1 through 3 (Fig. 1).
Platelet Counts and Risk of VTE
An elevated prechemotherapy platelet count was significantly associated with an increased rate of VTE (Fig. 2A, P for trend = 0.005). When the study population was divided by quartiles of prechemotherapy platelet counts, 3.6% of patients in the highest quartile (> 337,000/mm3) developed VTE, and this was significantly higher than the rate of 1.2% of patients who developed VTE in the lowest quartile (< 217,000/mm3, P = 0.0001). We analyzed this data further using approximation of quartile platelet counts to numbers that were more clinically relevant, and we observed a similar trend. Six hundred fifty-three (21.9%) patients had a platelet count of 350,000/mm3 or greater before starting chemotherapy on study. The incidence of VTE was 3.98% (1.66%/mo) for these patients, significantly higher than the rate of 1.25% (0.52%/mo) for patients with prechemotherapy platelet count of < 200,000/mm3 (P for trend=0.0003). This incidence was also significantly higher than the rate of 1.38% (0.58%/mo) observed for the rest of the study population (P for trend = 0.0001). The distribution of prechemotherapy platelet counts in patients who subsequently developed VTE was significantly higher than those who did not (Fig. 2B, t-test P = 0.002, Wilcoxon rank sum test P = 0.0002). Indeed, of 167 patients with prechemotherapy platelet counts < 150,000/mm3, none developed VTE. The increased risk of VTE with higher platelet counts persisted while the patients were on chemotherapy. Patients who developed VTE had significantly elevated mean platelet counts before each cycle of chemotherapy when compared with patients who did not develop VTE (P = 0.001) (Fig. 2C). Nadir platelet counts were also higher in patients who developed VTE (212,000 ± 15,333/mm3), compared with patients without VTE (159,000 ± 1466/mm3, P = 0.001).
Other Risk Factors for VTE
The primary site of cancer also affected the risk of VTE (P = 0.01), with the highest rates observed in patients with upper gastrointestinal cancers (including gastric, pancreatic, and hepatobiliary) (2.3%/mo), lung cancer (1.2%/mo), and lymphoma (1.1%/mo) (Table 2). Patients with a baseline hemoglobin < 10g/dL or who were receiving red-cell or white-cell growth factors during their first cycle of chemotherapy were also at increased risk of VTE. Patients with Eastern Cooperative Oncology Group (ECOG) performance status of 2 or greater had a trend toward an increased frequency of VTE, but this was not statistically significant (P = 0.21). The stage of cancer and particular types of chemotherapy regimens were also not significantly associated with VTE (P > 0.05).
|Characteristic||Rate of venous thromboembolism (%)||P value|
|All patients||58 (1.93)|
|Age < 65 yrs||37 (2.01)||0.72|
|Age ≥ 65 yrs||21 (1.82)|
|Stage 1||3 (0.78)|
|Stage 2||17 (2.22)|
|Stage 4||19 (1.77)|
|ECOG performance status|
|Site of cancer||0.0012|
|Upper gastrointestinal||5 (5.62)|
|Hodgkin disease||4 (8.16)|
|Non-Hodgkin lymphoma||4 (1.50)|
|Congestive heart failure||1 (1.45)||0.76|
|Myocardial infarction||2 (1.96)||0.99|
|Peripheral vascular disease||0 (0.00)||0.23|
|Cerebrovascular disease||3 (5.88)||0.04|
|Moderate or severe renal disease||2 (6.06)||0.08|
|Chronic pulmonary disease||4 (1.70)||0.78|
|History of surgery||20 (2.09)||0.68|
|BSA ≥ 2||16 (1.99)||0.9|
|BSA < 2||42 (1.92)|
|Prechemotherapy hematologic parameters|
|Platelet count ≥ 350,000/mm3||26 (3.98)||< 0.0001|
|Platelet count < 350,000/mm3||32 (1.37)|
|Hemoglobin < 10g/dL||10 (5.59)||0.0003|
|Hemoglobin ≥10g/dL||48 (1.71)|
|Use of growth factors during Cycle 1|
|White cell growth factors||28 (2.78)||0.02|
|Red cell growth factors||25 (3.16)||0.003|
|Type of regimen|
|Other cancer treatment|
|Hormonal therapy||5 (2.36)||0.64|
We combined cancer categories into upper gastrointestinal, lung, lymphoma, and other in developing a multivariate model for the risk of developing thromboembolism. In separate models, site of cancer was included as a categorical variable as well as an ordinal variable, with similar odds ratios estimated. In a multivariate logistic regression analysis, we found a strong association between hemoglobin < 10g/dL and use of red-cell growth factors (P < 0.0001), and we developed a combined variable for this. In the final multivariate model, variables found to be significantly and independently associated with development of symptomatic VTE included primary site of cancer (upper gastrointestinal or lung), a prechemotherapy platelet count ≥ 350,000/mm3, hemoglobin < 10g/dL or use of erythropoietin, and use of white-cell growth factors (Table 3). We found a significant first-order interaction between site of cancer and use of white-cell growth factors (P = 0.02). Patients with sites of cancer associated with higher risk of VTE (upper gastrointestinal, lung, or lymphoma) had a significantly increased risk of VTE associated with white-cell growth factor use (VTE rate of 5.9% vs. 1.52% without growth factor use, P = 0.0001; odds ratio 4.0, [95% CI, 1.8–8.7]). In contrast, patients with other sites of cancer did not appear to have an increased risk of VTE with the use of white-cell growth factors (VTE rate of 1.31% vs. 1.42% without growth factor use, P = 0.84).
|Patient characteristic||Odds ratio (95% CI)||P value|
|Site of cancera||0.03|
|Upper gastrointestinal||3.88b (1.43–10.05)||0.0076|
|Prechemotherapy platelet count ≥ 350,000/mm3||2.81 (1.63–4.93)||0.0002|
|Use of white cell growth factorsa||2.09 (1.21–3.61)||0.008|
|Hemoglobin < 10 g/dL or use of red cell growth factors||1.83 (1.07–3.14)||0.03|
Cancer patients are known to be at increased risk for VTE, but the general cancer population is heterogeneous and comprises newly diagnosed patients, those receiving active therapy, hospitalized patients, and those receiving end-of-life care. We chose to study the incidence of symptomatic VTE and its associated risk factors in ambulatory cancer patients who were beginning a new chemotherapy regimen in a prospective observational study. Unlike hospitalized patients, thromboprophylaxis is not considered standard of care for this subgroup of cancer patients. We found the frequency of VTE to be greater in our study population than previously estimated for the general cancer population. An elevated prechemotherapy platelet count was significantly and independently associated with increased risk of VTE. Certain cancer sites, particularly upper gastrointestinal, lung, and lymphoma, contributed disproportionately to the burden of VTE in this population.
The incidence of VTE observed in our study is higher than previously estimated for the general cancer population.10 It is difficult to directly compare the results from our study with previously published reports, however, because we chose a population of ambulatory patients on active treatment, but with a varied distribution of age, type and stage of cancer, and chemotherapy regimens. A study of risk factors associated with this population eliminates the heterogeneity observed in population-based or hospital-based studies of VTE in cancer, because patients that are immediately postoperative, suffering from debilitating comorbid illnesses, bedridden, or terminal would not be considered eligible for initiation of a new outpatient chemotherapy regimen. In a prospective study of metastatic breast cancer patients on chemotherapy, the rate of VTE observed in patients not on thromboprophylaxis was 4.4% over 188 days (0.7%/mo).26 This is similar to the rate of 1.49% over 2.4 months (0.62%/mo) observed in breast cancer patients in our study. In a retrospective study of cancer patients on chemotherapy, the annual incidence of symptomatic VTE was 10.9% (0.9%/mo), which is also similar to the 0.8%/month rate reported for our entire study population.29
We found a strong association between elevated platelet counts and risk of VTE during chemotherapy. Thrombocytosis in patients with essential thrombocythemia is associated with thrombotic events, but it has not been studied as a risk factor for VTE in large epidemiologic or prospective studies in the general population.4, 30, 31 A recent retrospective study of medical inpatients identified an admission platelet count of > 350,000/mm3 as predictive for VTE during hospitalization.32 The risk of VTE associated with an elevated platelet count in this study was similar in magnitude to that reported in our analysis (adjusted OR 3.1, 95% CI, 1.4–7).32 The association of platelet count with thrombosis may simply reflect an inflammatory state induced by a more advanced tumor. However, in our analysis, this association was independent of stage of cancer. It is possible, therefore, that platelets may play a greater role in the pathogenesis of cancer-associated VTE than previously known. Thrombocytosis is often observed in cancer patients,33 and interactions between platelet P-selectin and circulating carcinoma mucins have been described as a possible explanation for Trousseau syndrome.34 The poor prognosis described in association with thrombocytosis in various solid tumors35–38 may, at least in part, be explained by an increased rate of VTE and its attendant complications, as shown in our study. Further studies investigating the contribution of platelets to VTE in cancer are warranted.
The other risk factors for VTE observed in our study are consistent with prior reports. The significance of cancer site suggests that intrinsic factors unique to particular tumors contribute to the thrombophilic state. The high risk of VTE observed in patients with gastrointestinal or lung cancer is well known.21, 22, 39, 40 The association of lymphoma with VTE is increasingly being recognized.41–43 The presence of anemia or the use of red-cell growth factors was associated with an elevated risk of VTE, and this is consistent with earlier reports as well.44 Hematopoietic growth factors may be thrombogenic, although a recent metaanalysis was inconclusive.45, 46 In our study, white-cell growth factors were associated with VTE only in patients with cancers that had already predisposed to VTE (upper gastrointestinal, lung, and lymphoma). White-cell growth factors are often prescribed in the setting of infection, a known risk factor for VTE,30, 47, 48 and this may have contributed to the increased frequency observed. However, we found no association between documented infection during chemotherapy and the subsequent occurrence of VTE in our study population.
Our analysis had some limitations. The ANC Study Group Registry was designed to assess febrile neutropenia and related complications, as its primary endpoint, and not occurrence of VTE. However, data was prospectively collected, and the sample size was adequate for an analysis of this nature. We identified patients with symptomatic VTE, which is the most clinically significant endpoint. It is likely that there were additional patients with subclinical disease, thereby underestimating the true rate. The diagnosis of VTE also was not standardized but depended on usual clinical practice. Our analysis could not include biologic variables such as expression of tissue factor by tumor cells, or plasma levels of various hemostatic factors that may be relevant to the pathogenesis of VTE in this setting. Certain cancers known to be strongly associated with VTE, such as brain tumors, were underrepresented in the population.
In summary, this prospective observational study describes the frequency of symptomatic VTE in cancer patients initiating chemotherapy and suggests that patients with upper gastrointestinal or lung cancers, elevated prechemotherapy platelet count, anemia, and on growth factor treatment are at greatest risk for developing VTE during chemotherapy. Future directions include development of a clinical predictive model based on this analysis, and subsequent validation in a cohort of approximately 1500 patients that will be accrued in a second phase of this ongoing prospective observational study. Once validated, such a predictive model can help identify a subgroup of ambulatory patients at higher risk for developing VTE. Currently, thromboprophylaxis is not recommended for ambulatory cancer patients, despite a randomized trial showing its effectiveness in breast cancer patients.24, 26 Future clinical trials could evaluate the effectiveness and feasibility of thromboprophylaxis in high-risk subgroups of cancer patients.
- 2Lectures on clinical medicine. London: R. Hardwicke, 1867., , .
- 19Thromboembolism in hospitalized neutropenic cancer patients. J Clin Oncol. In press., , , , .
- 23Increased thromboembolic complications with concurrent tamoxifen and chemotherapy in a randomized trial of adjuvant therapy for women with breast cancer. National Cancer Institute of Canada Clinical Trials Group Breast Cancer Site Group. J Clin Oncol. 1996; 14: 2731–2737., , , , , .
- 43Venous Thromboembolic events in patients with diffuse large B cell non- Hodgkin's lymphoma [abstract]. Blood. 2004; 104: 4542., , , , , , .