Cancer and venous thromboembolism Andrea Piccioli, MD, a Paolo Prandoni, MD, a Bruce M. Ewenstein, MD, b and Samuel Z. Goldhaber, MD b Padua, Italy, and Boston, Mass. Trousseau 1 was the first researcher to suggest an association between cancer and hypercoagulability when he observed episodic migratory thrombophlebitis among patients with cancer. Thrombotic complications are especially frequent in patients with mucin-secreting adenocarcinomas, brain tumors, and certain hematologic malignancies such as acute promyelocytic leukemia and myeloproliferative disorders. Moreover, during the first 6 months after the diagnosis of deep vein thrombosis (DVT), previously occult cancer is often detected. 2 The appearance of migratory thrombophlebitis, nonbacterial thrombotic endocarditis, or thromboses in unusual sites such as the upper extremity, portal, or hepatic veins, is of particular concern. Abnormal platelet-vessel wall interactions and enhanced secondary hemostasis predispose patients who have cancer to thrombosis. These abnormalities m a y be related, directly or indirectly, to the tumor or may arise as complications of treatment. In some instances, the systemic effects of cancer affect the entire vasculature. In other circumstances, the cancer may induce local hemostatic defects that result in thromboses anatomically related to the tumor. This article reviews the relation between cancer and venous thromboembolism (VTE) and highlights some relevant clinical implications. PATHOGENESIS Pathogenetic mechanisms accounting for the development of thrombotic disorders in patients affected by cancer were described by Virchow more From %he Istituto Di Semeiotica Medica, University of Padua, and bthe Department of Medicine, Brigham and Women's Hospital, Harvard Medical School, Boston. Dr. Goldhaber receives support from the National Heart, Lung, and Blood Institute Academic Award in Systemic and Vascular Medicine (HL 02663). Received for publication Dec. 7, 1995; accepted Jan. 10, 1996. Reprint requests: Samuel Z. Goldhaber, MD, Cardiovascular Division, Brigham and Women's Hospital, 75 Francis St., Boston, ~ 02115. Am Heart J 1996;132:850~5. Copyright © 1996 by MosbY-Year Book, Inc. 4/1/74444 0002-8703/96/$5.00 + 0 850 than a century ago. They include hypercoagulability due to tumor cell activation of clotting, vessel wall injury, and stasis. Several platelet abnormalities have also been recognized. Hyperaoagulability. Neoplastic cells can activate the clotting system directly, thereby generating thrombin. Tumor cells can also stimulate the clotting system indirectly. For example, some tumor cells can bind independently to factors Va and Xa, thus enhancing the rate of prothrombinase complex formation and permitting further generation of thrombin. Direct tumor-cell activation of clotting. Tissue factor or similar thromboplastins can activate the extrinsic pathway through interaction with factor VIIa. Tissue factor procoagulant activity has been identified in tumors and tumor cell lines derived from h u m a n beings and animals. 3 It has been described in some acute leukemias 4 and in solid tumors of the stomach, ovary, and kidney. 5 Direct factor X activation with the procoagulant cysteine proteinase has been found in some patients with lung, prostate, colon, breast, and kidney cancer and with leukemia. 6, 7 Mucin-secreting adenocarcinomas are frequently associated with thrombosis because the sialic acid moiety can cause nonenzymatic activation of factor X to its active form, factor Xa. Consequently, adenocarcinomas of the lung, pancreas, gastrointestinal tract, and ovary are often associated with venous thrombosis, s Indirect tumor-cell activation of clotting. Tumor cells can activate systemic coagulation by stimulating mononuclear cells to synthesize and express various procoagulant substances, including tissue factor and factor X activators. Peritoneal exudate cells and tumor-associated macrophages in rabbits bearing the V2 carcinoma express increased tissue factor activity. 9 Normal monocytes and macrophages can be activated by tumor cells in the presence of lymphocytes. 1° In patients with cancer, endothelial cells m a y be activated by cytokines such as tumor necrosis factor and interleukin-1 or interleukinlike substances that may induce tissue factor production.11 A peptide produced by a h u m a n bladder cancer cell line Volume 132, Number 4 American Heart Journal stimulates tissue factor expression in endothelial cells. ]2 Tumor cells can secrete a peptide, vascular permeability factor, which can account for the increased microvascular permeability found in a variety of tumors. ]3 Vascular permeability factor also induces monocyte activation and chemotaxis across collagen membranes and an endothelial cell monolayer. 14 Therefore, systemic activation of coagulation may also reflect diffusion into plasma of extravascular clotting factors derived from solid-tumor growth. 15 The enhanced clotting activation in patients with cancer is confirmed by the demonstration of increased levels of systemic hypercoagnlability markers, such as fibrinopeptide A (FPA), prothrombin fragment F1.2, and thrombin-antithrombin III complexes in most patients. 16 Alternatively, clinical manifestations of increased thrombin generation may be accentuated by the failure of counterregulatory mechanisms. For example, in patients with extensive liver involvement, there may be decreased hepatic synthesis of antithrombin III and protein C. Other patients with cancer may have consumption coagulopathy or endothelial cell dysfunction that contribute to a hypercoagnlable state. Laboratory evaluation. The combination of increased thrombin generation and down-regulation of endothelial cell counterregulatory mechanisms leads to a prothrombotic state. Often this condition takes the form of chronic and well-compensated disseminated intravascular coagulation, which is characterized by subtle laboratory features. Routine laboratory tests such as those for prothrombin time and activated partial thromboplastin time (aPTT) may actually be shortened because of increased levels of circulating activated coagulation factors. The fibrinogen level is typically normal or slightly increased, as is the platelet count. Additional tests, including those for FPA, thrombin fragment F1.2, and thrombin-antithrombin III complexes may be useful adjuncts in the diagnosis of a hypercoagnlable state. Fibrinopeptide A. FPA is a peptide released by thrombin's proteolytic action on plasma fibrinogen and thus is a marker of thrombin generation. This marker is very sensitive for detecting intravascular thrombin but is not specific for intravascular thrombosis, because increased FPA levels m a y derive from extravascular sites or localized microvascular thrombosis as a consequence of infection, inflammation, or traumatic venipuncture. Increased FPA levels accompany many types of tumor, signifying increased thrombin generation. FPAlevels appear to be greater in patients with disseminated malignancy than in those with locally limited disease.!7 Piccioli et al. 851 Prothromb in fragment F ~.2. Prothrombin fragment F1.2 is a peptide released from the amino terminus of the thrombin moleculel during the conversion from prothrombin to thrombin, by factor Xa cleavage. Hanzal and Tatra is followed 76 patients with various gynecologic malignancies and 25 healthy women. Two thirds of the patients with malignant disease showed increased plasma F1.2 concentrations, whereas F1.2 levels remained normal among the control group without cancer, is Thrombin-antithrombin III complexes. Thrombin-antithrombin III complexes are stable enzymeinhibitor complexes derived from the linkage between thrombin and its endogenous heparan sulfate inhibitor antithrombin III. Increased concentrations of thrombin-antithrombin III complexes indicate increased thrombin generation and have often been found in patients with cancer. 19, 20 Vessel wall injury. There is increasing awareness that cancer cells can injure endothelium. In one study, the adhesion of tumor cells to endothelinm was evaluated in vivo by means of suspensions of transplantable thymic lymphoma and Walker 256 carcinoma that were injected into the inferior vena cava of mice and rats, respectively. 21 When the endothelial lining was not injured, platelet tumor aggregates adhered to intact endothelium. An observed separation of endothelial cell intercellular junctions gave tumor cells access to a highly thrombogenic subendothelia] surface. If the endothelial lining was mechanically injured, malignant cells adhered to a monolayer of platelets and fibrin in the subendothelium. Both patterns indicated a complex interaction among endothelium, platelets, and tumor cells. 21 Drug toxicity from chemotherapy can in some cases be traced directly to vascular endothelial cell damage (Table I). 22"26 Nicolson and Custead 27 described three types of drugs that can induce vascular damage: (1) drugs that can cause immediate effects on vascular endothelial cell integrity, such as bleomycin, carmustine (1,3-bis[2-chloroethyl)-l-nitrosourea [BCNU]), and vincristine; (2) drugs that can cause delayed effects on vascular endothelial cell integrity, such as adriamycin; and (3) drugs that have no apparent effects on vascular endothelium, such as fluorouridine. Combinations of cyclophosphamide, methotrexate, and 5-fluorouracil may cause reversible decreases in plasma levels of protein C and protein S. 2s, 29 Children with acute lymphoblastic leukemia have been considered at high risk for thromboembolic events when undergoing L-asparaginase chemotherapy. After administration of L-asparaginase, plasma October 1996 852 Piccioli et al. AmericanHeartJournal Table I. Risk of venous and arterial thrombosis in patients with cancer during chemotherapy without prophylaxis DVT Authors, year No. of patients Type of cancer During chemotherapy After chemotherapy Weiss et al., 22 1981 Goodnough et al.,23 1984 Levine eta]., 24 1988 Saphner et al.,25 1991 Levine et al.,26 1994 433 159 205 2352 159 Breast, stage II Breast, stage IV Breast, stage II Breast, various stages Breast, stage IV 22 (5%) 24 (15%) 14 (7%) 128 (5%) 7 (4%) 0 4 (2%) 0 0 0 inhibition of thrombin decreases because of a decrease in plasma concentration of inhibitors, especially antithrombin III. 3° Venous stasis. Venous stasis--caused by immobility, venous obstruction, increased venous pressure, venous dilatation, and increased blood viscosity-predisposes to VTE by preventing activated coagulation factors from being diluted and cleared by normal blood flow. Furthermore, hypoxic damage to endothelial cells due to stasis may produce prothrombotic alterations, including reduced levels of Surface-bound thrombomodulin and increased expression of tissue factor. In some patients with cancer, the degree of hypoxia is compounded by severe anemia or cardiopulmonary disease. Immobility in severely debilitated patients with cancer or in conjunction with cancer surgery predisposes to venous thrombosis because it promotes blood pooling into the intramuscular venous sinuses of the calf. Venous obstruction due to extrinsic vascular compression may develop in patients with bulky tumor masses. In addition, some tumors, most notably renal cell carcinoma, display a propensity to invade vascular structures, thereby producing intraluminal obstruction and disruption of the normal endothelial lining. Platelet abnormalities. Under normal physiologic conditions, the endothelial cell lining of the blood vessel provides a thromboresistant surface that inhibits platelet adherence a n d activation. In cancer patients, the normal function of endothelium may be disrupted through a variety of quantitative or qualitative defects in platelet function that predispose to VTE. Reactive thrombocytosis is frequently found in patients with cancer, especially those with advanced disease of the lung, colon, stomach, ovary, or breast. 31 Thrombocytosis may be caused by spontaneous clumping of platelets in the blood or by increased levels of thrombopoietin, a plasma glycoprotein that directly regulates the differentiation and maturation of megakaryocytes. Increased thrombopoietic activ- ity has been found in the serum of some patients with cancer and thrombocytosis. 31-33 Thrombocytosis is also a common and in some cases diagnostic feature of the myeloproliferative syndromes. In these disorders, dysregulation of thrombopoiesis arises from a clonal stem cell abnormality that produces platelets characterized by functional abnormalities. Many solid tumors demonstrate increased platelet adhesiveness and increased platelet activation. Usually, increased platelet activation results from increased generation of platelet agonists, principally thrombin and adenosine diphosphate. Thrombocytopenia is also common among patients with malignancies. It may reflect impaired generation of platelets, often brought about by irradiation or administration of chemotherapy, or it can result from marrow invasion by tumor cells. Thrombocytopenia in these patients also can be caused by increased platelet consumption or by accelerated destruction or sequestration of platelets in enlarged spleens. 34 CLINICAL IMPLICATIONS Indications for prophylaxis Primary prophylaxis. Because VTE is often encountered in patients with cancer, some clinicians have proposed that all patients with cancer should receive pharmacologic prophylaxis. However, further trials are needed before this approach can be endorsed. Currently, primary prevention should be considered for cancer patients in certain circumstances such as during and immediately after chemotherapy. A recent study demonstrated the advantage of using warfarin 1 mg/day for the prevention of thrombotic events in patients with breast cancer. 26 This approach appears to be cost effective. 35 Primary prevention also can be useful when central venous lines are placed. Bern et al.36 demonstrated an advantage to using warfarin 1 mg/day to prevent upper-extremity venous thrombosis. Finally, when the risk of VTE Volume 132, Number 4 American Heart Journal Piccioli et al. Table II. Risk of postoperative DVT in patients with cancer undergoing general surgery Table III. Incidence of subsequent cancer detection in patients with idiopathic and secondary VTE Frequency of postoperative DVT Authors, year With cancer Frequency of cancer Withoat cancer Authors, year Kakkar et al., 37 1970 Walsh et al., 3s 1974 Rosenberg et a l . y 1975 Sue-Ling et al., 4° 1986 Allan e t a ] . , z~l 1983 Sasahara, 42 1984 Total 24/59 (41%) 16/45 (35%) 28/66 (42%) 12/23 31/100 9/37 120/330 (52%) (31%) (22%) (36%) 38/144 (26%) 22/217 (10%) 29/128 (23%) 16/62 21/100 13/53 139/704 (26%) (21%) (24%) (20%) is increased by prolonged immobilization, trauma, or surgery, primary prevention may be of benefit. In patients with cancer, the risk of developing of postoperative DVT is greater than in patients with other disorders (Table I I ) Y -42 Moreover, postoperative pulmonary embolism in cancer patients is much more frequent than in patients without cancer. 43 Secondary prophylaxis. "Treatment" of venous thrombosis usually consists ofanticoagulation and can be more precisely described as "secondary prophylaxis" against recurrent VTE. For those patients who have pelvic or lower-extremity venous thrombosis but who are unable to tolerate anticoagulation, placement of an inferior vena caval filter should be considered. As is appropriate, patients with cancer who have venous thrombosis rarely receive primary therapy with thrombolysis. The limited cases in which thrombolysis m a y be considered include extension of venous thrombosis despite intensive anticoagulation; upperextremity thrombosis in patients who have an indwelling central venous catheter, which must be kept patent; or superior vena caval syndrome that does not improve with anticoagulation, arm elevation, or (if a tumor is present) radiation therapy. Thrombolysis is contraindicated in patients with cancer who may have tumor in the brain or spinal cord and in those with bulky disease in the chest, abdomen, or pelvis. Whenever possible, heparin should be administered as soon as a there is a reasonable possibility that venous thrombosis exists. Heparin is a much more effective anticoagulant than warfarin in patients with cancer. Ordinarily, warfarin can be initiated after a therapeutic aPTT (at least twice the control value) has been documented. The warfarin dose should be adjusted according to the inteimational normalized ratio (INR), which can be targeted to 3.0. Higher INRs are needed for patients in whom 853 Aderka et al., 46 Monreal et al., 47 1988 Monreal et al., 4s 1991 Prandoni et al., 49 1992 Years of In idiopathic In secondary follow-up DVT DVT 3.0 0.7 1.1 2.0 12/35 8/21 7/31 11/145 (34%) (38%) (23%) (8%) 2/48 (4%) 1/73 (1%) 5/82 (6%) 2/105 (2%) recurrent venous thrombosis develops despite this level of anticoagulation. In some patients, the thrombosis is completely resistant to warfarin, particularly in patients with venous clot in more than one extremity. These patients should be treated with long-term high-dose heparin, which can be given in one of several ways. One approach involves the use of adjusted subcutaneous unfractionated heparin administered three times daily. However, this strategy is time consuming, uncomfortable for the patient because of multiple injections, and yields highly variable aPTTs because of the erratic absorption and bioavailability of unfractionated heparin. 44 An alternative approach is to use low-molecularweight heparin with once- or twice-daily subcutaneous injections. If this latter strategy is undertaken, plasma heparin levels should be used if it is necessary to adjust the dose of low-molecular-weight heparin. A reasonable objective is a target plasma heparin level in the range of 0.3 to 0.7 units/ml measured 4 to 6 hours after injection. A third strategy is to arrange for a continuous infusion of unfractionated heparin on an ambulatory basis, particularly for patients receiving home hospice care. Adjustment of heparin concentrations is greatly facilitated with this approach, and patients are spared the discomfort of multiple daily injections of heparin. A population-based study in Rochester, Minnesota found that among patients receiving anticoagulation, the presence of cancer was significantly associated with major hemorrhage and recurrent thromboembolism. Cancer was more of a risk factor for hemorrhage and VTE recurrence than was old age. 45 Finally, the optimal duration of oral anticoagulation therapy after a first episode of VTE is unknown. Long-term anticoagulation may be worthwhile until it is certain that the cancer is cured. However, this approach must be weighed against the increased cu- October 1996 854 AmericanHeartJournal Piccioli et al, mulative bleeding risk, particularly bleeding into sites of tumor. Risk of subsequent cancer in patients with VTE. VTE may be the first indication of malignancy in an otherwise healthy patient (Table III). 46-49 There is a statistically significant and clinically important association between idiopathic VTE and the development of overt cancer, especially in patients with recurrent VTE. A retrospective cohort study among consecutive patients with a symptomatic idiopathic DVT showed a 12% incidence of cancer detected after an initial screening at referral. During a minimum of 18 months of follow-up, only 3% of 112 DVT patients were newly diagnosed with cancer.5° Among patients with VTE due to an unidentified cause, a diagnostic examination for occult malignancy according to standard clinical practice is worthwhile. However, extensive screening, such as total-body computed tomography or magnetic resonance imaging, cannot be recommended because available data indicate that few new cancers are identified this way and that the ultimate effect on survival probably is trivial. Nevertheless, further investigation is needed to assess the utility of extensive screening for neoplasms in patients with idiopathic VTE. SUMMARY Neoplastic cells can activate the clotting system directly, thereby generating thrombin, or indirectly, by stimulating mononuclear cells to synthesize and express various procoagulants. Clinical manifestations of increased thrombin generation may be accentuated by down-regulation of endothelial cell counterregulatory mechanisms, such as decreased hepatic synthesis of antithrombin III and protein C. Cancer cells and chemotherapeutic agents can injure endothelial cells, thereby intensifying hypercoagulability. In addition, normal endothelial cell function may be disrupted by various defects in platelet function. Currently, primary prevention of venous thrombosis should be considered for cancer patients (1) during and immediately after chemotherapy, (2) when long-term indwelling central venous catheters are placed; or (3) when hospitalization for cancer is characterized by prolonged immobilization, trauma, or surgery. Secondary prevention of recurrent venous thrombosis usually necessitates anticoagulation. In some patients with cancer, the condition is resistant to warfarin, and long-term adjusted high-dose heparin is required. For patients unable to tolerate heparin or warfarin because of major bleeding problems, placement of an inferior vena caval filter should be considered. The diagnosis of venous thrombosis may help to uncover previously occult carcinoma by prompting a complete physical examination, chest roentgenography, and mammography. However, extensive cancer screening with total-body computed tomography or magnetic resonance imaging has not been shown to be cost effective for patients with venous thrombosis. REFERENCES 1. Trousseau A. Phlegmasia alba dolens. Clinique medicale de l'hotel-dieu de Paris. Paris: J-B Balliere et Fils, 1864;3:654-712. 2. Nordstrsm M, Lindblad B, Anderson H, Bergqvist D, KjellstrSm T. Deep venous thrombosis and occult malignancy: an epidemiological study. BMJ 1994;398:89I-4. 3. Francis JL, Carvalho M, Francis DA. The clinical value of tissue factor assays. Blood Coagul Fibrinolysis 1995;6:$37-44. 4. Kubota T, Andoh K, Sadakata H, Tanaka H, Kobayashi N. Tissue factor released from leukemic cells. Thromb Haemost 1991;65:59-63. 5. Francis JL. Haemostasis and cancer. Med Lab Sci 1989;46:331-46. 6. Falanga A, Alessio MG, Donati MB, Barbni T. A new procoagulant in acute leukemia. Blood 1988;71:870-5. 7. Gordon SG, Benson B. Analysis of serum cancer procoagulant activity and its possible use as a tumor marker. Thromb Res 1989;56: 431-40. 8. Scares SM. Diagnosis and treatment of cancer-related thrombosis. Semin Thromb Hemost 1992;18:373-9. 9. Lorenzet R, Peri G, Locati D, Allavena P, Colucci M, Semeraro N, et al. Generation of procoagnlant activity by mononuclear phagocytes: a possible mechanism contributing to blood clotting activation within malignant tissues. Blood 1983;62:271-3. 10. Lando PA, Edgington TS. An innate host response to the neoplastic cell: syngeneic rat tumor cells can elict a rapid de novo lymphoid procoagulant response. J Immunol 1985;135:3587-95. 11. Miyauchi S, Moroyama T, Kyoizumi S, Asakawa J-I, Okamoto T, Takada I~ Malignant tumor celllines produce inter].eukin-l-like factor. In Vitro Cell Dev Biol 1988;24:753-8. 12. Nog~chi M, Sakai T, Kisiel W. Identification and partial purification of a novel tumor-derived protein that induces tissue factor on cultured human endothelial cells. Biochem Biophys Res Commun 1989; 160:222-7. 13, Dvorak HF, Nagy JA, Berse B, Brown LF, Yeo KT, Dvorak AM, et al, Vascular permeability factor, fibrin, and the pathogenesis of tumor stroma formation. Ann N Y Acad Sci 1992;667:101-1L 14. Clauss M, Gerlach M, Gerlach H, Brett J, Wang F, Familletti PC, et al. Vascular permeability factor: a tumor-derived polypeptide that induces endothelial cell and monocyte procoagnlant activity, and promotes monocyte migration. J Exp Med 1990;172:1535-45. 15. Dvorak HF. Abnormalities of hemostasis in malignant disease. In: Colman RW, Hirsh J, Marder VJ, Salzman EW, editors. Hemostasis and thrombosis: basic principles and clinical practice. 3rd ed. Philadelphia: JB Lippincott, 1994:1238-54. 16. Rickles FR' Edwards RL" Leuk°cytes and tum°r cells in thr°mb°sis"In: Colman RW, Hirsh J, Marder VJ, Salzman EW, editors. Hemostasis and thrombosis: basic principles and clinical practice. 3rd ed. Philadelphia: JB Lippincott, 1994:1164-79. 17. Lindahl AK' Sandset PM' Abildgaard U" Indices °f hyperc°agnlati°n in cancer as compared with those in acute inflammation and acute infarction. Haemostasis 1990;20:253-62. 18. Hanzal E, Tatra G. Prothrombin fragment FL2 plasma concentrations in patients with gynecologic malignancies. Gynecol Oncol 1993; 49:373-6. 19. Gabazza EC, Taguchi O, Yamakami T, Machishi M, Ibata H, Suzuki S. Evaluating prethrombotic state in lung cancer using molecular markers. Chest 1993;103:196-200. 20. Kemkes-Matthes B, Bleyl H. Factor IXi-antithrombin (IXIAT) and thrombin-antithrombin (TAT) complexes in lung cancer patients. Ann Hematol 1992;64:35-9. Volume 132, Number 4 Piccioli et al. American Heart Journal 21. Naschitz JE, Yeshurun D, Lev LM. Thromboembolism in cancer: changing trends. Cancer 1993;71:1384-90. 22. Weiss RB, Tormey DC, Holland JF, Weinberg VE. Venous thrombosis during multimodal treatment of primary breast carcinoma. Cancer Treat Rep 1981;65:677-9. 23. Goodnough LT, Saito H, Manni A, Jones PK, Pearson OH. Increased incidence of thromboembolism in stage IV breast cancer patients treated with a five-drug chemotherapy regimen: a study of 159 patients. Cancer 1984;54:1264-8. 24. Levine MN, Gent M, Hirsh J, Arnold A, Goodyear MD, Hry~liuk W, et al. The thrombogenic effect of anticancer drug therapy in women with stage II breast cancer. N Engl J Med 1988;318:404-7. 25. Saphner T, Tormey DC, Gray R. Venous and arterial tlirombosis in patients who received adjuvant therapy for breast cancer. J Clin Oncol 1991;9:286-94. 26. Levine M, Hirsh J, Gent M, Arnold A, Warr D, Falanga A, et al. Double-blind randomised trial era very-low-dose warfarin for prevention of thromboembolism in stage IV breast cancer. Lancet 1994;343:886-9. 27. Nicolson GL, Custead SE. Effects of chemotherapeutic drugs on platelet and metastatic tumor cell-endothelial cell interactions as a model for assessing vascular endothelial integrity. Cancer Res 1985;45:331-6. 28. Rogers JS II, Murgo AJ, Fontana JA, Raich PC. Chemotherapy for breast cancer decreases plasma protein C and protein S. J Clin Oncol 1988;6:276-81. 29. Kuzel T, Esparaz B, Green D, Kies M. Thrombogenicity of intravenous 5-fluorouracil alone or in combination with cisplatin. Cancer 1990; 65:885-9. 30. Mitchell L, Hoogendoorn H, Giles AR, Vegh P, Andrew M. Increased endogenous thrombin generation in children with acute lymphoblastic leukemia: risk of thrombotic complications in L-asparaginase-induced antithrombin III deficiency. Blood 1994;83:386-91. 31. Grignani G, Pacchiarini L, Pagliarlno M. The possible role of blood platelets in turnout growth and dissemination. Haematologica 1986; 71:245-55, 32. Rosenthal DS, Murphy S. Thrombecytosis. In: Handin RI, Lux SE, Stossel TP, editors. Blood: principles and practice of hematology. Philadelphia: JB Lippincott, 1995:439-55. 33. Corbett G, Perry DJ. Significance of thrombocytosis [letter]. Lancet 1983;1:77. 34. Ratnoff OD. Hemostatic emergencies in malignancy. Semin Oncol 1989;16:561-71. 35. Rajan R, Gafni A, Levine M, Hirsh J, Gent M. Very low-dose warfarin prophylaxis to prevent thromboembolism in women with metastatic breast cancer receiving chemotherapy: an economic evaluation. J Clin Oncol 1995;13:42-46. 36. Bern MM, Lokich JJ, Wallach SR, Bothe A Jr., Benotti PN, Arkin CF, 855 et al. Very low doses of warfarin can prevent thrombosis in central venous catheters: a randomized prospective trial. Ann Int Med 1990; 112:423-8. 37. Kakkar W , Howe CT, Nicolaides AN, Renney JTG, Clarke MB. Deep vein thrombosis of the leg: is there a high risk' group? Am J Surg 1970;120:527-30. 38. Walsh JJ, Bonnar J, Wright FW. A study of pulmonary embolism and deep leg vein thrombosis after major gynaecological surgery using labelled fibrinogen-phlebography and lung scanning. J Obstet Gynaecol Br Commonw 1974;81:311-6. 39. Rosenberg IL, Evans M, Pollock AV. Prophylaxis of postoperative leg vein thrombosis by low dose subcutaneous heparin or peroperative calf muscle stimulation: a controlled clinical trial. Br Med J 1975;1:649-51. 40. Sue-Ling HM, Johnston D, McMahon MJ, Philips PR, Davies JA. Preoperative identification of patients at high risk of deep venous thrombosis al~er elective major abdominal surgery. Lancet 1986;1:1173-6. 41. Allan A, Williams JT, Belton JP, Le Quesne LP. The use of graduated compression stockings in the prevention of postoperative deep vein thrombosis. Br J Surg 1983;70:172-4. 42. The Multicentre Trial Committee. Dihydroergotamine-hepariu prophylaxis of postoperative deep vein thrombosis: a multicentre trial. JAMA 1984;251:2960-6. 43. Huber O, Bounameaux H, Borst F, Rohner A. Postoperative pulmonary embolism aider hospital discharge: an underestimated risk. Arch Surg 1992;127:310-3. 44. HirschDR, LeeTH, MorrisonRB, CarlsonW, Goldhaber SZ. Shortened hospitalization by means of adjusted-dose subcutaneous heparin for deep venous thrombosis. Am Heart J 1996;131:276-80. 45. Gitter MJ, Jaeger TM, Pettersen TM, Gersh BJ, Silverstein MD. Bleeding and thromboembolism during anticoagulant therapy: a population-based study in Rochester, Minnesota. Mayo Clin Proc 1995; 70:725-33. 46. Aderka D, Brown A, Zelikovski A, Pinkhas J. Idiopathic deep vein thrombosis in an apparently healthy patient as a premonitory sign of occult cancer. Cancer 1986;57:1846-9. 47. Monreal M, Salvador R, Soriano V, Sabria M. Cancer and deep venous thrombosis [letter]. Arch Intern Med 1988;148:485. 48. Monreal M, Lafoz E, Casals A, Inaraja L, Montserrat E, Callejas JlYl, et al. Occult cancer in patients with deep venous thrombosis: a systematic approach. Cancer 1991;67:541~5. 49. Prandoni P, Lensing AWA, Bfiller HR, Cogo A, Prins MH, Cattelan AM, et al. Deep-vein thrombosis and the incidence of subsequent symptomatic cancer. N Engl J Med 1992;327:1128-33. 50. Comuz J, Pearson S, Creager MA, Goldman L. Initial evaluation for cancer in patients with symptomatic deep vein thrombosis [abstract]. J Invest Med 1995;43(Suppl 2):304A.