Ovarian cancer and venous thromboembolic risk

Transcription

1 Gynecologic Oncology 99 (2005) Ovarian cancer and venous thromboembolic risk S. Tateo a, *, L. Mereu b, S. Salamano a, C. Klersy c, M. Barone d, A.C. Spyropoulos e, F. Piovella d a Department of Obstetrics and Gynecology, IRCCS Policlinico San Matteo, Piazzale Golgi, Pavia, Italy b Department of Obstetrics and Gynecology, University of Pavia, Italy c Biometry and Clinical Epidemiology Service, IRCCS Policlinico San Matteo, Pavia, Italy d Thromboembolic Unit, IRCCS Policlinico San Matteo, Pavia, Italy e Clinical Thrombosis Center, Lovelace Medical Center, Albuquerque, NM 87108, USA Received 25 March 2005 Available online 28 June 2005 Abstract Objective. To determine the incidence and the prognostic factors of objectively diagnosed deep vein thrombosis (DVT) and pulmonary embolism (PE) in patients with epithelial ovarian malignancy. Methods. We reviewed the records of all patients with epithelial ovarian cancer who were diagnosed, treated, and followed-up at our institution between 1990 and Data were collected regarding age, body mass index, previous DVT and PE, menopause status, FIGO stage, grade, histology, type of surgery, residual disease, first line chemotherapy, and relapse status. Results. Of the 253 cases, the overall incidence of symptomatic venous thromboembolic events (VTE) was 16.6% (42 patients): 1.6% (4) with PE and 15% (38) with DVT. 8 events (3.2%) were detected before tumor diagnosis, 6 (2.4%) in the postoperative period, 16 (6.4%) during first line chemotherapy and 12 (4.8%) throughout the follow-up period. Risk factors associated with occurrence of VTE were: at diagnosis, history of deep vein thrombosis ( P = 0.001); during chemotherapy, older age ( P = 0.017), larger body mass index ( P = 0.019), FIGO stage 2c 4 ( P = 0.004), no surgery ( P = 0.003), and presence of residual tumor ( P = 0.026). None of the considered risk factors were found to be predictors of VTE postoperatively. The multivariate regression analysis found that residual tumor, age, and body mass index were independent prognostic factors. Conclusion. The incidence of VTE throughout the entire history of ovarian malignancy is high. Prognostic factors could be used to establish prophylaxis protocols based on risk stratification. D 2005 Elsevier Inc. All rights reserved. Keywords: Ovarian cancer; Deep vein thrombosis; Pulmonary embolism; Chemotherapy Introduction The association between cancer and venous thromboembolism (VTE) is well known. The pathogenic mechanism was described by Virchow with the triad of hypercoagulability, vessel wall injury, and stasis: a complex interaction between the tumor cell, the patient, and the hemostatic system. Neoplastic cells can activate the clotting system directly, via thrombin generation, and indirectly by stimulating * Corresponding author. Fax: address: oncologia.ginecologica@smatteo.pv.it (S. Tateo). mononuclear cells to produce and express pro-coagulant substances [1]. Cancer cells can also injure endothelium by direct vascular invasion and by secretion of vascular permeability factors which account for extravascular accumulation of fibrinogen around tumor growth [2]. Extrinsic factors such as chemotherapy, and venous catheters are also responsible for direct injury of the vessel wall [3]. Venous stasis predisposes to venous thrombosis by preventing activated coagulation factors from being diluted and cleared by normal blood flow. Hypoxic damage to endothelial cells due to stasis may produce prothrombotic alterations. Venous stasis develops as a /$ - see front matter D 2005 Elsevier Inc. All rights reserved. doi: /j.ygyno

2 120 S. Tateo et al. / Gynecologic Oncology 99 (2005) consequence of immobility in debilitated cancer patients, surgery, or as a result of venous obstruction due to extrinsic vascular compression in patients with bulky tumor masses [4]. Particularly, in patients with ovarian cancer, it has been shown that a hematocrit-independent hyperviscosity syndrome with elevated platelet count with the concurrence of increased coagulation activation, elevated plasma fibrinogen concentration, reduced red blood cells deformability, dehydration due to malignant ascites, and activation of the host inflammatory response, may alter the rheologic properties of blood and contribute to reduce blood flow. These facts may be associated with the spreading of disease and with a poor prognosis [5]. No study, so far, has provided evidence that any of the plasma markers of coagulation activation are indeed capable of identifying patients at risk of developing thrombotic complications. Because it is not possible to accurately predict who among patients with cancer will develop thrombosis, it appears reasonable to consider a strategy of routine anti-thromboprophylaxis in such patients. In order to verify this, it is essential to establish if the magnitude of VTE risk is sufficiently great, and if there are evaluable risk factors. There are few studies about the incidence of VTE in ovarian cancer patients [5 8] and most of them use nonobjective diagnostic methods that may overestimate the clinical problem. On these bases, we undertook the present study, in order to analyze the incidence and prognostic factors of deep vein thrombosis and/or pulmonary embolism in association with ovarian malignancy. Materials and methods Patient population We reviewed the data obtained by the charts of patients with epithelial ovarian cancer who were diagnosed, treated, and followed at San Matteo Hospital of Pavia from 1990 to Inclusion criteria included patients with histological or instrumental diagnosis of epithelial ovarian cancer (by total abdomen computed tomography or ultrasound) confirmed with cytology from ascitic fluid, that were treated with surgery and/or chemotherapy and followed-up for at least 1 month after first line chemotherapy, or until death. Exclusion criteria included patients with cachexia not suitable for chemotherapy and inadequate clinical data collection about surgery, chemotherapy, or follow-up. All patients that underwent surgery received antithrombotic prophylaxis as follows: from 1990 up to 1999, unfractionated heparin (5000 I.U. t.i.d.); from 2000 onwards, low molecular weight heparin (4.000 I.U. o.d.), given for at least 7 days starting the evening before surgery, or up to mobilization. All patients submitted to chemotherapy were contacted and followed-up weekly for 20 weeks, and subsequently every 3 months for the first 3 years, every 6 months for the next 3 years, and then yearly up to 10 years from cancer diagnosis. Data collection For each patient data on age at diagnosis, body mass index (BMI), previous history of deep vein thrombosis or pulmonary embolism, menopause status, FIGO stage, grade, histology subtype, type of surgery, residual disease, first line chemotherapy, relapse status, and outcome were collected. In particular, we considered 4 types of surgeries or procedures: (1) paracentesis for cytological evaluation of ascites; (2) explorative laparotomy with biopsy; (3) monoor bilateral salpingo-oophorectomy with or without total hysterectomy, appendicectomy, and intra- and extraperitoneal biopsies; (4) mono- or bilateral salpingo-oophorectomy with or without: total hysterectomy, appendicectomy, intraand extraperitoneal biopsies, omentectomy, pelvic paraaortic limphadenectomy, and debulking procedures. The residual tumor after surgery was classified as absent, <2 cm in diameter, and >2 cm in diameter. The first line chemotherapy schedules were: cysplatinum (CDDP 50 mg/mq weekly); cysplatinum + other drugs (CDDP 50 mg/mq + cyclophosphamide 600 mg/mq T adriamycin 45 mg/mq every 3 4 weeks); carboplatin (JM8 AUC6 every 3 4 weeks); carboplatin + other drugs (JM8 AUC 6 + paclitaxel mg/mq every 3 weeks, JM8 AUC6 + epirubicin 120 mg/mq every 4 weeks, JM8 AUC5 + paclitaxel 175 mg/mq + topotecam 1 mg/mq per 3 days every 3 weeks). Study outcome The outcome of the study was a combined endpoint of symptomatic DVT and PE. The diagnosis of deep vein thrombosis and pulmonary embolism was objectively assessed following clinical suspicion of an event by compression ultrasound (CUS) for lower limb deep vein thrombosis and spiral computed tomography for pulmonary embolism. We also separately considered events occurring: (1) at diagnosis or in the preceding 6 months (the thrombotic event being the first manifestation of the neoplasia); (2) postoperatively, from surgery to day 30 or to the beginning of chemotherapy; (3) during first line chemotherapy, from the first cycle until 4 weeks after the last cycle; and (4) during subsequent follow-up. Statistical analysis Mean and standard deviation (SD), or median and 25th 75th percentiles, and counts expressed as percentages (%) were computed to describe continuous and categorical variables, respectively. Logistic regression (or Fisher exact

3 S. Tateo et al. / Gynecologic Oncology 99 (2005) test) was used to assess the association of potential risk factors with the pre- and postoperative events, while Cox regression (or log rank test) was used for events occurring during chemotherapy or late follow-up. Odds ratio (OR), hazard ratio (HR), and 95% confidence intervals (95% CI) were reported. Finally, to consider the entire observation time as a whole, a multivariate Cox model for competing risks was fitted, in order to identify prognostic factors for thrombosis during postoperative, intra-chemotherapeutic, and long-term follow-up periods. Baseline risks were allowed to be different for each endpoint by using model stratification. Prognostic factors identified at univariate analysis were included in the multivariate model, after assessing for multicollinearity. Akaike Information Criterion (AIC) was used to select the best among competing models. Stata 8 (StataCorp, College Station, TX) was used for computation. A 2-sided P value of 0.05 was considered statistically significant. Results A total of 253 patients, aged 59.6 years (SD 12.5), with epithelial ovarian cancer were investigated. Twenty-three patients were not included in the study because of inadequate clinical data (surgery in other hospitals, lost to follow-up) or because of cachexia. One-hundred and thirty patients died over a median observation time of 31 months (25th 75th percentile 12 68), corresponding to a death rate of 15.2/100 person/years (95% CI ). Patient characteristics are shown in Table 1. Twenty-eight patients underwent perioperative antithrombotic prophylaxis with low molecular weight heparin and 217 with unfractionated heparin. Four patients had a vena caval filter inserted before surgery. Chemotherapy started a median of 18 days (25th 75th percentile ) after surgery. The median observation time from surgery to the first occurring event or to the end of follow-up for patients free of DVT and PE was 24.3 months (25th 75th percentile ). The cumulative incidence over the entire follow-up period of clinical thromboembolic events was 16.6% (42 patients), 1.6% (4) of these were PE and 15% (38) were DVT. Eight events (3.2%) were recognized at tumor diagnosis while the remaining 34 were observed after Notes to Table 1: Abbreviations: unknown X, body mass index BMI (m/kg 2 ), deep vein thrombosis DVT, performance status PS. a Residual tumor after surgery (cm). b Surgery: 0 = no surgery; 1 = explorative bioptic laparotomy; 2 = mono bilateral salpingo-oophorectomy with or without: total hysterectomy, appendicectomy, intra- and extraperitoneal biopsies; 3 = mono bilateral salpingo-oophorectomy with or without: total hysterectomy, appendicectomy, intra- and extraperitoneal biopsies, omentectomy, pelvic and paraaortic limphadenectomy, debulking. c Chemotherapy schedule: 0 = no chemotherapy, CDDP = cysplatinum; CDDP+ = cysplatinum + other (cyclophosphamide, adriamycin); JM8 = carboplatin; JM8+ = carboplatin + other (paclitaxel, epirubicin, topotecam). surgery; 6 (2.4%) at the time of surgery, 16 (6.4%) during first-line chemotherapy (2 of these occurring within 30 days from surgery) and the remaining 12 (4.8%) during long-term follow-up; of these, 5 (2%) occurred in patients with recurrent disease and 7 (2.8%) during further chemotherapy. Table 1 Patient characteristics Age (years) 59.8 (125; 21 85) BMI (kg/m 2 ) 24.6 (4.7; 17 37) Performance status, N (%) (87.3) 1 23 (9.1) 2 4 (1.6) 3 2 (0.8) 4 1 (0.4) X 2 (0.8) Prior DVT, N (%) 10 (3.9) Menopause, N (%) 193 (76.3) FIGO stage, N (%) 1a 51 (20.1) 1b 3 (1.2) 1c 22 (8.7) 2a 3 (1.2) 2b 5 (2.0) 2c 7 (2.8) 3a 3 (1.2) 3b 8 (3.1) 3c 121 (47.8) 4 29 (11.5) X 1 (0.4) Residual tumor a, N (%) (41.1) 2 35 (13.8) >2 111 (43.9) X 3 (1.2) Histology subtypes, N (%) Serous 119 (47.1) Mucinous 43 (17.0) Endometrioid 31 (12.3) Clear cell 17 (6.7) Indifferentiated 11 (4.3) Mix 1 (0.4) Other 20 (7.9) X 11 (4.3) Grading, N (%) 1 27 (10.7) 2 68 (26.9) (54.1) 4 10 (4.0) X 11 (4.3) Surgery b, N (%) 0 8 (3.2) 1 19 (7.4) 2 70 (27.7) (61.7) Chemotherapy c, N (%) 0 53 (21.0) CDDP 60 (23.7) CDDP+ 53 (21.0) JM8 11 (4.3) JM8+ 74 (29.2) Other 2 (0.8)

4 122 S. Tateo et al. / Gynecologic Oncology 99 (2005) Fig. 1 shows the TVP/EP free survival from surgery throughout follow-up. All deep vein thrombosis occurred in the lower limbs, 7 of them were bilateral. Patients characteristics according to the 4 event types are shown in Table 2, while their prognostic relevance are summarized in Table 3. Regression analysis revealed that a prior DVT was the only significant predictor of DVT/PE before diagnosis, while some increase in risk (although not statistically significant) was shown for PS (>0) and clear cell histology [2/17 vs. 5/225, OR 5.9 (95% CI ), P = 0.08]. None of the considered risk factors were found to be predictors of DVT/PE postoperatively. Older age, larger BMI, higher FIGO stage (2c to 4), and presence of residual tumor after surgery, all appeared to significantly increase the risk of DVT/PE during chemotherapy. Finally, the following risk factors were associated with the occurrence of DVT/PE during long-term subsequent follow-up: age, history of DVT, FIGO stage 2c 4, presence of residual tumor after surgery. For the multivariate analysis, encompassing the entire observation time, 3 different models were fitted due to the presence of multicollinearity between histology, stage and residual tumor, that included age, BMI, and each of the 3 tumor characteristics in turn. Based on the AIC, the later model was considered the best. Residual tumor (Hazard Ratio of 3.75) and 5 unit increase in BMI (Hazard Ratio of 1.44), significantly increased the risk of thrombosis in any setting (surgery, chemotherapy, and long-term follow up) (Table 4). Discussion Fig. 1. Kaplan Meier DVT/PE free survival from surgery. Venous thrombembolism is increasingly recognized as a common complication in patients with malignant disease. Trousseau in 1865 first described hypercoagulability and thrombosis in cancer. Further multiple clinical, pathologic, and laboratory studies support the notion that activation of coagulation is not simply an epiphenomenon but may be related to enhanced tumor growth, angiogenesis, and treatment modalities. Prior to the present study, there have been no studies evaluating the incidence of clinically symptomatic VTE events in patients with ovarian cancer. From this point of view, we have investigated symptomatic VTE in ovarian cancer patients in relation to patient characteristics, type of malignancy, and treatment selection. In our study, the VTE overall incidence was 16%. Post-tumor diagnosis treatment modalities, as surgery and chemotherapy, are commonly considered factors that increase clinical TEE. The observed rate of DVT among gynecologic oncologic surgery patients not receiving prophylaxis has varied between 17% and 38% [9,10]. In the Enoxaparin and Cancer (ENOXACAN) I study [11], DVT detected by venography occurred in 15% of patients receiving 10 days of prophylaxis after surgery for cancer; the ENOXACAN II study confirmed that DVT and PE can occur up to 4 weeks after major cancer surgery with a 4.8% event rate in patients receiving days of prophylaxis [12]. The present study found, until 30 days after surgery, a clinically symptomatic VTE incidence of 2.4% and 3.2% if we consider patients in postoperative period already undergoing chemotherapy. In fact, our study counted 2 of 16 DVT events during chemotherapy, occurring within 30 days from surgery. von Tempeloff found a venographic incidence of DVT of 10.6% in 60 ovarian cancer patients undergoing chemotherapy [6]. In our study, 16 patients (6.4%) had clinically overt DVT and PE during first-line chemotherapy. The mechanism of chemotherapy-induced thrombosis has been studied and experimental observations have been made by Levine et al.: decreased protein C, increased fibrinopeptide A production, increased endothelial cell reactivity, release of tissue factor from monocytes and endothelial cells, downregulation of thrombomodulin, and decreased fibrinolytic response [3]. von Tempeloff maintained that chemotherapy has a direct interaction with vascular endothelial or blood cells [7]. Previous population registries found that the standardized incidence ratio was highest within the first 6 months of an idiopathic DVT and subsequent identification of an occult malignancy [13,14]. We found that 8 (3.2%) patients with deep vein thrombosis occurred in the 6 months before the diagnosis of ovarian cancer and this result was statistically related to previous DVT with an OR of With regards to prognostic factors and VTE incidence in gynecological cancer surgery, Clarke Pearson et al. in a retrospective review of gynecologic cancer patients submitted to mechanical prophylaxis with intermittent pneumatic compression, found that diagnosis of cancer, history of deep vein thrombosis, and age greater than 60 years were independent prognostic factors and that patients with two or three of these variables had a 3.2% incidence of developing VTE compared with a 0.6% incidence in patients with no or one risk factor [15]. Prichard in

5 S. Tateo et al. / Gynecologic Oncology 99 (2005) Table 2 Distribution of patient characteristics in the presence of each type of DVT/PE Events Before diagnosis (8/253) Postoperatively (6/245) During chemotherapy (16/239) During long-term follow-up (12/223) Mean (SD) Mean (SD) Mean (SD) Mean (SD) Age (years) DVT/PE Yes 65.5 (10.4) 55.7 (13.7) 65.9 (12.3) 63.3 (10.3) DVT/PE No 59.6 (12.5) 59.7 (12.5) 59.2 (12.4) 59 (12.4) BMI (kg/m 2 ) DVT/PE Yes 26.6 (6.0) 22.7 (4.8) 27.9 (7.4) 26.9 (14.0) DVT/PE No 24.5 (4.7) 24.6 (4.7) 24.3 (4.4) 24.2 (4.4) N (%) N (%) N (%) N (%) PS 0 5/221 (2.3) 4/215 (1.8) 14/212 (6.6) 11/202 (5.4) /30 (10.0) 2/28 (7.1) 2/25 (8) 1/25 (4) X 0/2 (0) 0/2 (0) 0/2 (0) 0/2 (0) Prior DVT No 5/243 (2.1) 5/238 (2.1) 15/233 (6.4) 10/223 (4.4) Yes 3/10 (30.0) 1/7 (14.3) 1/6 (16.6) 2/6 (33.3) Menopause No 1/60 (1.7) 2/59 (3.4) 2/57 (3.5) 1/57 (1.7) Yes 7/193 (3.6) 4/186 (2.1) 14/182 (7.6) 11/172 (6.4) FIGO stage I IIb 1/84 (1.2) 3/82 (3.7) 1/80 (1.2) 2/82 (2.4) IIc IV 7/169 (4.9) 3/162 (1.8) 15/158 (9.4) 10/146 (6.8) X 0/1 (0) 0/1 (0) Histology Clear cell + mucinous 2/60 (3.3) 2/58 (3.4) 3/57 (5.2) 1/55 (1.8) Other 5/182 (2.7) 3/176 (1.7) 12/173 (6.9) 11/164 (6.7) X 1/11 (9) 1/11 (9) 1/9 (11.1) 0/9 (0) Grading 1 0/27 (0) 1/26 (3.8) 2/26 (7.6) 1/25 (4) indiff 7/215 (3.3) 4/208 (1.9) 13/204 (6.3) 11/194 (5.6) X 1/11 (9.0) 1/11 (9) 1/9 (11.1) 0/9 (0) Surgery No 2/8 (25.0) 1/7 (14.2) Yes 14/231 (6) 11/222 (4.9) Surgery (detailed) 0 1 0/26 (0) 2/25 (8.0) 0/23 (0) 2 2/68 (2.9) 0/67 (0) 2/55 (3.6) 3 4/151 (2.6) 14/49 (28.5) 10/151 (6.6) Residual tumor 0 3/101 (3.0) 3/99 (3) 1/99 (1.0) 2 1/32 (3.1) 3/31 (9.6) 4/29 (13.8) >2 2/101 (2.0) 10/106 (9.4) 7/98 (7.1) X 0/11 (0) 0/2 (0) 0/3 (0) Chemotherapy 0 0/50 (0) 1/53 (1.9) CDDP+ 11/108 (10.1) 4/99 (4) Other 5/81 (6.1) 7/77 (9) For abbreviations, see Table 1. univariate analisys found that increased weight and age were statistically significant prognostic factors for VTE in breast cancer patients during chemotherapy [16]. Our study revealed that the risk of DVT during first-line chemotherapy increases in patients with an older age, a larger BMI, a higher FIGO stage (2c to 4) and the presence of residual tumor after surgery. von Tempeloff confirmed that ovarian cancer patients with DVT arising during chemotherapy were postmenopausal, significantly older and with a higher BMI compared with patients without DVT [7]. In the same study, no correlation was observed between the development of DVT and tumor characteristics including histology type, grading, and FIGO stage. Lastly, von Tempeloff s study on 60 ovarian cancer patients did not find a correlation between prognostic factors and postoperative DVT [6]. The multivariate regression analysis in the present study found that residual tumor increased the risk of thrombosis by more

6 124 S. Tateo et al. / Gynecologic Oncology 99 (2005) Table 3 Evaluation of potential risk factors for each type of DVT/PE Events Before diagnosis Postoperatively During chemotherapy During long-term follow-up OR (95% CI) P value OR (95% CI) P value HR (95% CI) P value HR (95% CI) P value Age (per 10 years) 1.5 ( ) ( ) ( ) ( ) BMI (per 5 units) 1.4 ( ) ( ) ( ) ( ) PS ( ) ( ) ( ) (0.6 40) Prior DVT No Yes 20.4 ( ) ( ) ( ) (1.7 37) Menopause No Yes 2.2 ( ) ( ) ( ) (0.7 41) # FIGO stage I IIb IIc IV 3.6 ( ) ( ) (1.1 65) (1.5 34) Histology Other Clear cell/mucinous 1.2 ( ) ( ) ( ) ( ) # Grading indiff NE* 1.000* 0.5 ( ) ( ) (0.2 15) Surgery No 1 1 Yes 0.22 ( ) # 0.11 ( ) # Surgery (detailed) # # 2 1 NE # NE # 3 NE 1.000* NE #, NE #, - Residual tumor No Yes 0.7 ( ) ( ) ( ) <0.001 Chemotherapy No 1 Yes 4.3 (0.6 31) 0.145^ Chemo (detailed) 0 Cddp+ 1 Other 0.6 ( ) For abbreviations, see Table 1. NE = not valuable. * Fisher exact test. # Log rank test. ^ Time-dependent Cox model; surgery 2 vs. 3. P = P = than 3-fold, and BMI by 1.5-fold per 5 unit increase, although we acknowledge the limitations of this analysis in situations with a small number of outcome events. Nevertheless, these data confirm the one revealed in the univariate analysis: that residual tumor may be a new prognostic factor for VTE in patients with ovarian cancer. Lastly, in Table 4 Multivariate analysis (Cox model for competing risks) encompassing postoperative, chemotherapy, and long-term follow-up DVT/PE [Model v 2 (3) = 21.85, P ] HR (95% CI) P value Residual tumor 3.57 ( ) Age (per 10 years increase) 1.37 ( ) BMI (per 5 units increase) 1.44 ( ) <0.001 accordance with von Tempeloff [6], we did not find any correlation between postoperative DVT and any of the prognostic factors that we analyzed, probably due to the use of effective thromboprophylaxis. We have to underline that this is a retrospective study with a limited number of patients and potential bias. Some variables are not shown to be significant factors maybe because our analysis is not adequately powered to detect meaningful differences. Prophylaxis of DVT is cost effective in terms of live-years gained even for patients with relatively short life expectancies as ovarian cancer patients [17]. As such, the results of present study may be used in the development of optimal strategies of thromboprophylaxis in patients with ovarian cancer by helping to define VTE time of occurrence and prognostic factors for this group of patients.

7 S. Tateo et al. / Gynecologic Oncology 99 (2005) References [1] Prandoni P, Piccioli A, Girolami A. Cancer and venous thromboembolism: an overview. Haematologica 1999;84: [2] Edwards RL, Rickles FR. Thrombosis and cancer. In: Hull R, Pineo GF, editors. Disorders of thrombosis. Philadelphia WB Saunders Co.; p [3] Levine MN. Prevention of thrombotic disorders in cancer patients undergoing chemotherapy. Thromb Haemostasis 1997;78: [4] Salzman EX, Hirsh J. The epidemiology, pathogenesis, and natural history of venous thrombosis. In: Colman EW, Hirsh J, Marder VJ, Salzman EW, editors. Hemostasis and thrombosis: basic principles and clinical practice. Philadelphia J.B. Lippincott Co.; p [5] von Tempelhoff GF, Heilmann L, Hommel G, Schneider D, Niemann F, Zoller H. Hyperviscosity syndrome in patients with ovarian carcinoma. Cancer 1998;82: [6] von Tempelhoff GF, Dietrich M, Niemann F, Schneider D, Hommel G, Heilmann L. Blood coagulation and thrombosis in patients with ovarian malignancy. Thromb Haemostasis 1997;77: [7] von Tempelhoff GF, Niemann F, Schneider D, Kirkpatrick CJ, Hommel G, Heilmann L. Blood rheology during chemotherapy in patients with ovarian cancer. Thromb Res 1998;90: [8] Canney PA, Wilkinson PM. Pulmonary embolism in patients receiving chemotherapy for advanced ovarian cancer. Eur J Cancer Clin Oncol 1985;21: [9] Clarke-Pearson DL, Synan IS, Colemen RE, Hinshaw W, Creasman WT. The natural history of postoperative venous thromboemboli in gynaecologic oncology: a prospective study of 382 patients. Am J Obstet Gynecol 1984;148: [10] Crandon AJ, Koutts J. Incidence of post-operative deep vein thrombosis in gynaecologic oncology. Aust NZJ Obstet Gynecol 1983;23: [11] ENOXACAN study group. Efficacy and safety of enoxaparin versus unfractionated heparin for prevention on deep vein thrombosis in elective cancer surgery: a double blind randomized multicentre trial with venographic assessment. Br J Surg 1997;84: [12] Bergqvist D, Agnelli G, Cohen AT, Eldor A, Nilson PE, Le Moigne- Amrani A, et al. Duration of prophylaxis against venous thromboembolism with enoxaparin after surgery for cancer. N Engl J Med 2002;346: [13] Baron JA, Gridley G, Weiderpass E, Nyren O, Linet M. Venous thromboembolism and cancer. Lancet 1998;351: [14] Søorensen HT, Mellemkjær L, Steffensen FH, Olsen JH, Nielsen GL. The risk of a diagnosis of cancer after primary deep venous thrombosis or pulmonary embolism. N Engl J Med 1998;338: [15] Clarke-Pearson DL, Dodge RK, Synan I, McClelland RC, Maxwell GL. Venous thromboembolism prophylaxis: patients at high risk to fail intermittent pneumatic compression. Obstet Gynecol 2003;101: [16] Prichard KI, Paterson AHG, Paul NA, Zee B, Fine S, Pater J. Increased thromboembolic complications with concurrent tamoxifen and chemotherapy in a randomized trial of adjuvant therapy for women with breast cancer. J Clin Oncol 1996;14: [17] Maxwell GL, Myers ER, Clarke-Pearson DL. Cost-effectiveness of deep vein thrombosis in gynaecologic oncology surgery. Obstet Gynecol 2000;95: