the James P. Wilmot Cancer Center and the Departments of Medicine and Biostatistics and Computational Biology, University of Rochester, Rochester, NY.

    ABSTRACT

    PURPOSE: Cancer is associated with thrombosis, but the frequency of thromboembolism in hospitalized cancer patients receiving current chemotherapy regimens is not known. We investigated venous and arterial thromboembolism and associated outcomes in hospitalized cancer patients actively receiving therapy, as identified by neutropenia.

    METHODS: We conducted a retrospective cohort study using the discharge database of the University HealthSystem Consortium. This included 66,106 adult neutropenic cancer patients with 88,074 hospitalizations between 1995 and 2002 at 115 medical centers in the United States.

    RESULTS: Thromboembolism was reported in 5,272 patients (8%), with 5.4% patients developing venous thromboembolism and 1.5% developing arterial thromboembolism during the first hospitalization. Patients with lymphoma and leukemia accounted for one third of venous and nearly one half of arterial events. Clinical variables most frequently associated with thromboembolism were age  65 years; primary site of cancer, including lung, GI, gynecologic, and brain; and comorbidities, including infection, pulmonary and renal disease, and obesity. In-hospital mortality was significantly greater in patients with venous (odds ratio [OR] = 2.01; 95% CI, 1.83 to 2.22) or arterial thromboembolism (OR = 5.04; 95% CI, 4.38 to 5.79). From 1995 to 2002, there was a 36% increase in venous events and a 124% increase in arterial events (P < .0001 for trend).

    CONCLUSION: Thromboembolism is frequent in hospitalized neutropenic cancer patients, including in perceived low-risk subgroups such as patients with hematologic malignancies and nonmetastatic disease, and seems to be increasing. Thromboembolism is associated with increased in-hospital mortality. Increased efforts at thromboprophylaxis are warranted.

    INTRODUCTION

    An association between thrombosis and malignancy has been reported since at least the 19th century, and the risk of venous thromboembolism is increased in cancer patients, especially in those receiving chemotherapy.1-3 The estimated annual incidence of thromboembolism in the cancer population is 0.5%,4 and thromboembolism is the second leading cause of death in these patients.5 An analysis of Medicare claims data for hospital discharges from 1988 to 1990 indicated that cancer patients developed venous thromboembolism during initial hospitalization significantly more frequently than noncancer patients, although incidence rates were only 0.6% and 0.57%, respectively.6 More recently, 7.8% of cancer patients treated at three medical centers developed venous thromboembolism over a median follow-up time of 26 months.7

    Varying rates may be observed because of the heterogeneity of the hospitalized cancer population, which includes patients with a new diagnosis of cancer and patients admitted subsequently for surgery, elective chemotherapy, complications of therapy, or end-of-life care. The incidence of thromboembolism among these subgroups may vary, and the implications for prophylaxis and therapy differ.8 Therefore, we chose to study cancer patients with neutropenia as reflecting a large and more homogenous population of patients receiving active therapy. We investigated the proportion of patients with venous and arterial thromboembolism in this population and characterized its association with in-hospital mortality and other variables.

    METHODS

    All discharge summaries of adult cancer patients with febrile neutropenia admitted between 1995 and 2002 to one of 115 medical centers in the United States were reviewed using the discharge database of the University HealthSystem Consortium. Patients were identified using International Classification of Diseases, 9th revision, clinical modification (ICD-9-CM) codes that contained at least one diagnosis of malignant disease (ICD-9-CM 140 to 208) and the diagnosis of agranulocytosis (ICD-9-CM 288). Patients with thromboembolism were identified using codes for venous thrombosis (codes 451.0 to 451.9, 452, 453.0 to 453.9, and 557.0), pulmonary embolism (codes 415.1 to 415.9), arterial embolism (codes 444.0 to 444.9), acute cerebrovascular disease (codes 433.0 to 434.9 and 436), and acute coronary arterial disease (codes 410.0 to 410.9 and 411.1 to 411.8). Patients with a history of venous thromboembolism or of receiving anticoagulation were identified using codes V125.1, V125.2, and V586.1. The major categories of cancer that were studied (and their codes) included lung (162 and 163), breast (174), colon (153), rectum (154), stomach (151), pancreas (157), other abdominal cancers (152, 155, 156, 158, and 159), ovary (183), endometrium and cervix (179 to 182), head and neck (140 to 149), esophageal (150), Hodgkin's disease (201), non-Hodgkin's lymphoma (200 and 202), brain (191 and 192), prostate (185), bladder (188), renal (189), leukemia (204 to 208), testicular (186), myeloma (203), and sarcoma (170 and 171). Comorbidities and risk factors included documented infection (001 to 139.8, 480 to 486, and 9966.2), pulmonary disease (487 to 519.9), hypertension (401), renal disease (580 to 593.9), diabetes mellitus (250 to 250.9), tobacco abuse (305.1 to 305.12 and V1582), congestive heart failure (428 to 428.9), hepatic disease (570 to 576.9), and obesity (278). These comorbidities and risk factors referred to the period of hospitalization only. In-hospital mortality was defined as death from any cause during the period of hospitalization.

    Statistical Analysis

    The association of thromboembolism with clinical variables was studied using a univariate analysis and reported as odds ratios (ORs) with 95% CIs. The association results were re-evaluated using a multivariate logistic regression model with venous thromboembolism as the response variable. The full cohort of 66,106 patients was used in the multivariate model because data regarding age, sex, and mortality were complete (except for data on 90 patients regarding mortality). For missing data regarding race, an other/unknown category was created. Clinically relevant factors, including age  65 years, race, sex, comorbidities, and sites of cancer known to be associated with thromboembolism, were included in the model. To test homogeneity of ORs between mortality and thromboembolism, we split the data set into 54 groups based on zip code and year (P = .20 for venous thromboembolism and P = .10 for arterial thromboembolism). The association between mortality and thromboembolism was also tested in a multivariate analysis. The association of categorical variables with outcomes were based on a 2 test while the Cochran-Armitage test was used to determine trend. Statistical significance of independent variables in multivariate analyses were based on the Wald 2 test. Statistical analysis was conducted using SAS version 8.2 (SAS Institute, Cary, NC).

    RESULTS

    Patient Characteristics

    A total of 88,074 hospitalizations with neutropenia occurred in 66,106 cancer patients between 1995 and 2002 in 115 medical centers, including 14,727 patients (22.3%) with multiple hospitalizations. The average age was 53 years, and 26.2% of patients were greater than 65 years old (Table 1). Nearly three fourths of the population (72.8%) was white, with blacks representing 11.2% and Hispanics representing 4.6% of the population. Hematologic malignancies were common and included leukemia (22.1%), non-Hodgkin's lymphoma (19.6%), myeloma (6.8%), and Hodgkin's disease (3.1%). Solid tumors were categorized into 17 types, and these represented one third of the population. A small minority of patients (1.4%) had multiple cancers.

    Thromboembolic Events

    Thromboembolic events were reported in 5,272 patients (7.98%), including 4,434 patients (6.71%) who developed thromboembolism during the first hospitalization (Table 2). Of 14,727 patients with multiple hospitalizations, 838 (5.69%) developed thromboembolism during subsequent hospitalizations. For further analysis, we considered events during the first hospitalization only. The most frequent events were venous thrombosis (4.79%) and pulmonary embolism (0.93%). Arterial events, including acute coronary disease (0.8%), acute cerebrovascular disease (0.5%), and arterial embolism (0.2%), occurred less frequently. A prior history of venous thrombosis, pulmonary embolism, or chronic anticoagulation was more common in patients who developed venous thromboembolism than in patients who did not (P < .0001), although this comprised only a small percentage of patients (1.7%).

    The proportion of patients with venous and arterial thromboembolism increased by 36% and 124%, respectively, over the 8 years of study (Fig 1; P < .0001 for trend for both). This increase occurred despite a decrease in median length of hospital stay from 8 to 7 days and a decline in the proportion of patients with lengths of stay greater than 10 days from 46.1% to 42.5% (P < .0001). Notably, the proportion of patients greater than 65 years old increased during the same period from 23.2% to 28.1%, as did the proportion of obese patients, which increased from 0.48% to 1.14% (P < .0001 for both).

    Sites of cancer with the highest proportion of patients with venous thromboembolism were pancreas (12.1%), brain (9.5%), and endometrial or cervical (9%; Table 3). Patients with non-Hodgkin's lymphoma (n = 650) and leukemia (n = 641) represented more than one third of all patients with venous events, and lung (n = 326) and breast (n = 204) cancers accounted for nearly one sixth of all patients with venous events. Sites of cancer with the highest proportion of patients with arterial thromboembolism were prostate (3.9%), lung (2.8%), and bladder (2.8%; Table 3). Patients with non-Hodgkin's lymphoma (n = 173) and leukemia (n = 279) represented nearly half of all patients with arterial events, with lung (n = 131) and breast cancer (n = 36) accounting for a further one sixth of patients.

    Variables significantly associated with venous thromboembolism included age  65 years; site of cancer, including pancreas, brain, endometrial or cervical, and lung (P  .01 for each); and presence of comorbidities, including infection, pulmonary disease, hypertension, renal disease, congestive heart failure, hepatic disease, and obesity (Table 1). Variables associated with arterial thromboembolism included age  65 years; male sex; black race; site of cancer, including leukemia, colon, prostate, and lung (P  .01 for each); and presence of comorbidities, including infection, pulmonary disease, hypertension, renal disease, diabetes mellitus, congestive heart failure, and hepatic disease. There was a significant association between the occurrence of venous and arterial thromboembolism (OR = 1.73; 95% CI, 1.38 to 2.16). Among patients with age  65 years, women were more likely to develop venous thromboembolism (5.7% of men v 6.6% of women; OR = 1.16; 95% CI, 1.02 to 1.31; P = .02), and the proportion was highest among black women with age  65 years (7.7% v 6.5% for other women aged  65 years; P = .13). Information regarding presence or absence of metastatic disease was available for only a subgroup of the population (n = 38,010). In this subgroup, venous thromboembolism was more frequent in patients with metastatic disease (OR = 1.23; 95% CI, 1.13 to 1.34), whereas arterial thromboembolism was not (OR = 0.59; 95% CI, 0.51 to 0.69).

    In-Hospital Mortality

    The in-hospital mortality for the entire population was 8.3% over the 8 years of study. Patients with venous thromboembolism had greater mortality than patients without such a diagnosis (OR = 2.01; 95% CI, 1.83 to 2.22; P < .0001), and a similar trend was observed in the subgroup of patients with information regarding metastatic disease (Fig 2A). In-hospital mortality was greater in patients with venous thromboembolism and either nonmetastatic disease (OR = 1.62; 95% CI, 1.37 to 1.91; P < .0001) or metastatic disease (OR = 2.06; 95% CI, 1.74 to 2.44; P < .0001). Similarly, in-hospital mortality was greater in patients with arterial thromboembolism (OR = 5.05; 95% CI, 4.38 to 5.79) and either nonmetastatic disease (OR = 3.6; 95% CI, 2.96 to 4.37; P < .0001) or metastatic disease (OR = 3.87; 95% CI, 2.93 to 5.13; P < .0001; Fig 2B).

    Multivariate Analysis

    The following clinical variables were found to be significantly associated with venous thromboembolism using a multivariate logistic regression analysis: age  65 years; site of cancer, including brain, lung, stomach, pancreas, other abdominal, ovary, endometrium, and cervix; arterial thromboembolism; and presence of comorbidities, including infection, pulmonary and renal disease, and obesity. Their association with venous thromboembolism is described in Table 4. A similar multivariate analysis for mortality identified venous (OR = 1.5; 95% CI, 1.3 to 1.6) and arterial thromboembolism (OR = 2.7; 95% CI, 2.3 to 3.2) as significantly associated with increased in-hospital mortality. Other variables of significance in this model were year of hospitalization (with mortality declining over the duration of study), age  65 years, race, site of cancer, and comorbidities.

    DISCUSSION

    We studied the occurrence of thromboembolism in hospitalized cancer patients receiving chemotherapy, as identified by the presence of neutropenia. The analysis shows that venous and arterial thromboembolic events are frequent complications of hospitalization in this population and that they both contribute to increased in-hospital mortality. The proportion of patients with thromboembolism is high across all subgroups studied including patients with hematologic malignancies and with nonmetastatic disease, who are commonly perceived to be at lower risk for thromboembolism. Older age, particular sites of cancer, infection, and other comorbidities are associated with a higher risk of thromboembolic events. Of particular concern is the increasing frequency of venous and arterial thromboembolism.

    The proportion of patients with venous thromboembolism reported in our study is greater than the proportion observed in a prospective study of acutely ill medical patients, which found an incidence of 0.7% symptomatic venous thrombosis and 1% pulmonary embolism in the placebo arm.9 The proportion is also much higher than the 0.6% incidence observed in Medicare claims data from 1988 to 1990.6 The incidence of arterial thromboembolism in hospitalized cancer patients has not been previously reported. Several factors may have contributed to the high proportion of thromboembolism observed. The study population comprised only patients receiving chemotherapy because a discharge diagnosis of neutropenia was required. Nearly 44% of patients experienced an infectious complication during hospitalization. Both chemotherapy and infection are risk factors for thrombosis.3,10-12 Hematopoietic growth factors, which are commonly used in hospitalized patients with febrile neutropenia, may be thrombogenic, although a recent meta-analysis was inconclusive.13 The true risk of thromboembolism may have increased since the last report; indeed, we observed a significant increase in its occurrence over the study duration. There may be an increased awareness of the diagnosis of thromboembolism, leading to increased testing.14 Risk factors, such as patient age and obesity, increased over the same period, although length of hospitalization declined. Newer, more thrombogenic cancer treatment regimens, increasing use of implanted venous access devices, and an increased awareness of thrombosis as a complication of hospitalization may also have contributed.15-17 Although the discharge codes we used to identify venous thrombosis included superficial thrombophlebitis, the number of patients with this diagnosis alone was small (n = 198, 0.2%).

    The significance of cancer site suggests that intrinsic factors unique to these particular tumors contribute to hypercoagulability. The high risk of venous thromboembolism in patients with pancreas, brain, and gynecologic cancers is consistent with prior observations.7,18-20 The association of arterial thromboembolism with prostate, lung, and colon cancers and leukemia is a novel finding and deserves further investigation.

    The occurrence of thromboembolism has several clinical consequences related to patient morbidity, interruption of chemotherapy, and cost of hospitalization.21 In our analysis, thromboembolism was also associated with increased in-hospital mortality. Arterial thromboembolism and pulmonary embolism can be life threatening and could have contributed to the observed increase in mortality. However, venous thrombosis without embolism is not a life-threatening condition. Experimental models suggest that activation of the coagulation cascade confers a growth advantage on the tumor. In particular, tissue factor, thrombin, and fibrin(ogen) can enhance tumor growth, metastasis, and angiogenesis.22-25 The presence of venous thrombosis could signify an aggressive tumor biology and, therefore, a worse short-term prognosis. Cancer diagnosed at the same time as or within 1 year of an episode of venous thromboembolism has previously been shown to be associated with an advanced stage and a three-fold lower survival at 1 year.26

    Our study had several limitations. It was retrospective and based on discharge codes. However, accuracy of coding for pulmonary embolism and deep venous thrombosis has been previously shown to be 92% and 84%, respectively.27 Neither the time to thromboembolic event nor the time from thromboembolism to mortality were available. The study included only inpatients and may have underestimated the true proportion because venous thromboembolism is often diagnosed and managed on an outpatient basis. Although neutropenia is usually a manifestation of chemotherapy administration, it can also occur without the use of chemotherapy, particularly in patients with myeloproliferative disorders. In this study, however, the proportion of patients with thromboembolism was not substantially altered when patients with myeloproliferative disorders were excluded (data not shown). We could not validate the diagnosis of neutropenia in our study population, although it has been previously shown to be specific in similar administrative datasets.28,29 Thromboprophylaxis can alter the incidence of venous thromboembolism, but its use in our study population was not known.30 According to a recent survey, however, less than 5% of medical oncology patients are routinely administered thromboprophylaxis.31 Venous thromboembolism occurred in 5.4% of patients, and therefore, logistic regression analysis may be hampered by the rare disease assumption. Information regarding stage of disease, use of implanted intravascular access devices, type and timing of chemotherapy regimens, and outside-hospital mortality was not available. Documentation regarding other known risk factors for thromboembolism was also missing (eg, information on use of postmenopausal hormonal replacement therapy was available for < 0.1% of the study population).

    Despite these limitations, it is clear from our analysis that the risk of both venous and arterial thromboembolism is high and increasing in hospitalized neutropenic cancer patients, and the occurrence of thromboembolism portends a poor short-term prognosis. Thromboprophylaxis with warfarin and low molecular weight heparins has been shown to be effective in cancer outpatients and hospitalized acutely ill medical patients, respectively.9,30 These compounds may have other beneficial effects in cancer patients.32,33 Prospective studies are necessary to characterize the risks and benefits of thromboprophylaxis in this patient population.

    Authors' Disclosures of Potential Conflicts of Interest

    The authors indicated no potential conflicts of interest.

    Author Contributions

    Conception and design: Alok A. Khorana, Charles W. Francis, Gary H. Lyman

    Administrative support: Gary H. Lyman

    Collection and assembly of data: Eva Culakova, Gary H. Lyman

    Data analysis and interpretation: Alok A. Khorana, Charles W. Francis, Eva Culakova, Richard I. Fisher, Nicole M. Kuderer, Gary H. Lyman

    Manuscript writing: Alok A. Khorana, Charles W. Francis, Eva Culakova, Richard I. Fisher, Nicole M. Kuderer, Gary H. Lyman

    Final approval of manuscript: Alok A. Khorana, Charles W. Francis, Eva Culakova, Richard I. Fisher, Nicole M. Kuderer, Gary H. Lyman

    Acknowledgment

    We thank David Oakes, PhD, for critical comments and Susan Sullivan and Barbara Hartzog for editorial assistance.

    NOTES

    Supported by the James P. Wilmot Cancer Research Fellowship (A.A.K.) and National Institutes of Health Grant No. 2T32 ES007271 (E.C.).

    Presented in part at the 40th Annual Meeting of the American Society of Clinical Oncology, New Orleans, LA, June 5-8, 2004.

    Authors' disclosures of potential conflicts of interest and author contributions are found at the end of this article.

    REFERENCES

    Trousseau A, Bazire PV, Cormack JR: Lectures on Clinical Medicine. London, United Kingdom, R. Hardwicke, 1867

    Khorana AA: Malignancy, thrombosis and Trousseau: The case for an eponym. J Thromb Haemost 1:2463-2465, 2003

    Heit JA, Silverstein MD, Mohr DN, et al: Risk factors for deep vein thrombosis and pulmonary embolism: A population-based case-control study. Arch Intern Med 160:809-815, 2000

    Lee AY, Levine MN: Venous thromboembolism and cancer: Risks and outcomes. Circulation 107:I17-I21, 2003

    Ambrus JL, Ambrus CM, Mink IB, et al: Causes of death in cancer patients. J Med 6:61-64, 1975

    Levitan N, Dowlati A, Remick SC, et al: Rates of initial and recurrent thromboembolic disease among patients with malignancy versus those without malignancy: Risk analysis using Medicare claims data. Medicine (Baltimore) 78:285-291, 1999

    Sallah S, Wan JY, Nguyen NP: Venous thrombosis in patients with solid tumors: Determination of frequency and characteristics. Thromb Haemost 87:575-579, 2002

    Rickles FR, Levine MN: Epidemiology of thrombosis in cancer. Acta Haematol 106:6-12, 2001

    Samama MM, Cohen AT, Darmon JY, et al: A comparison of enoxaparin with placebo for the prevention of venous thromboembolism in acutely ill medical patients: Prophylaxis in medical patients with enoxaparin study group. N Engl J Med 341:793-800, 1999

    Bryant AE: Biology and pathogenesis of thrombosis and procoagulant activity in invasive infections caused by group A streptococci and Clostridium perfringens. Clin Microbiol Rev 16:451-462, 2003

    Keller TT, Mairuhu AT, de Kruif MD, et al: Infections and endothelial cells. Cardiovasc Res 60:40-48, 2003

    Levi M, Keller TT, van Gorp E, et al: Infection and inflammation and the coagulation system. Cardiovasc Res 60:26-39, 2003

    Barbui T, Finazzi G, Grassi A, et al: Thrombosis in cancer patients treated with hematopoietic growth factors: A meta-analysis—On behalf of the Subcommittee on Haemostasis and Malignancy of the Scientific and Standardization Committee of the ISTH. Thromb Haemost 75:368-371, 1996

    Stein PD, Hull RD, Kayali F, et al: Venous thromboembolism according to age: The impact of an aging population. Arch Intern Med 164:2260-2265, 2004

    Solit DB, Morris M, Slovin S, et al: Clinical experience with intravenous estramustine phosphate, paclitaxel, and carboplatin in patients with castrate, metastatic prostate adenocarcinoma. Cancer 98:1842-1848, 2003

    Zangari M, Barlogie B, Thertulien R, et al: Thalidomide and deep vein thrombosis in multiple myeloma: Risk factors and effect on survival. Clin Lymphoma 4:32-35, 2003

    Stein PD, Hull RD, Ghali WA, et al: Tracking the uptake of evidence: Two decades of hospital practice trends for diagnosing deep vein thrombosis and pulmonary embolism. Arch Intern Med 163:1213-1219, 2003

    Dhami MS, Bona RD, Calogero JA, et al: Venous thromboembolism and high grade gliomas. Thromb Haemost 70:393-396, 1993

    Wun T, Law L, Harvey D, et al: Increased incidence of symptomatic venous thrombosis in patients with cervical carcinoma treated with concurrent chemotherapy, radiation, and erythropoietin. Cancer 98:1514-1520, 2003

    von Tempelhoff GF, Dietrich M, Niemann F, et al: Blood coagulation and thrombosis in patients with ovarian malignancy. Thromb Haemost 77:456-461, 1997

    Elting LS, Escalante CP, Cooksley C, et al: Outcomes and cost of deep venous thrombosis among patients with cancer. Arch Intern Med 164:1653-1661, 2004

    Rickles FR, Patierno S, Fernandez PM: Tissue factor, thrombin, and cancer. Chest 124:58S-68S, 2003

    Sahni A, Sporn LA, Francis CW: Potentiation of endothelial cell proliferation by fibrin(ogen)-bound fibroblast growth factor-2. J Biol Chem 274:14936-14941, 1999

    Palumbo JS, Kombrinck KW, Drew AF, et al: Fibrinogen is an important determinant of the metastatic potential of circulating tumor cells. Blood 96:3302-3309, 2000

    Falanga A, Donati MB: Pathogenesis of thrombosis in patients with malignancy. Int J Hematol 73:137-144, 2001

    Sorensen HT, Mellemkjaer L, Olsen JH, et al: Prognosis of cancers associated with venous thromboembolism. N Engl J Med 343:1846-1850, 2000

    Kniffin WD Jr, Baron JA, Barrett J, et al: The epidemiology of diagnosed pulmonary embolism and deep venous thrombosis in the elderly. Arch Intern Med 154:861-866, 1994

    Madsen K, Rosenman M, Hui S, et al: Value of electronic data for model validation and refinement: Bacteremia risk in children with fever and neutropenia. J Pediatr Hematol Oncol 24:256-262, 2002

    Rosenman M, Madsen K, Hui S, et al: Modeling administrative outcomes in fever and neutropenia: Clinical variables significantly influence length of stay and hospital charges. J Pediatr Hematol Oncol 24:263-268, 2002

    Levine M, Hirsh J, Gent M, et al: Double-blind randomised trial of a very-low-dose warfarin for prevention of thromboembolism in stage IV breast cancer. Lancet 343:886-889, 1994

    Kakkar AK, Levine M, Pinedo HM, et al: Venous thrombosis in cancer patients: Insights from the FRONTLINE survey. Oncologist 8:381-388, 2003

    Zacharski LR, Henderson WG, Rickles FR, et al: Effect of warfarin anticoagulation on survival in carcinoma of the lung, colon, head and neck, and prostate: Final report of VA Cooperative Study #75. Cancer 53:2046-2052, 1984

    Khorana AA, Sahni A, Altland OD, et al: Heparin inhibition of endothelial cell proliferation and organization is dependent on molecular weight. Arterioscler Thromb Vasc Biol 23:2110-2115, 2003

查询更多Cancer相关信息在本站>>

《临床肿瘤学医学期刊》2006年1月第24卷第1期