Prevalence and Significance of Asymptomatic Venous Thromboembolic Disease Found on Oncologic Staging CT

CT of Cancer Patients with Asymptomatic Venous Thromboembolic Disease Chest Imaging Original Research Carmel G. Cronin 1 Derek G. Lohan 1 Maccon Keane 2 Clare Roche 1 Joseph M. Murphy 1 Cronin CG, Lohan DG, Keane M, Roche C, Murphy JM Keywords: cancer staging, CT staging, deep venous thrombosis, oncologic imaging, pulmonary embolism, thromboembolic disease, venous thromboembolism DOI:10.2214/AJR.07.2067 Received February 20, 2007; accepted after revision March 26, 2007. 1 Department of Radiology, University College Hospital, Newcastle Rd., Galway, Ireland. Address correspondence to C. G. Cronin (carmelcronin2000@hotmail.com). 2 Department of Oncology, University College Hospital, Galway, Ireland. AJR 2007; 189:162 170 0361 803X/07/1891 162 American Roentgen Ray Society Prevalence and Significance of Asymptomatic Venous Thromboembolic Disease Found on Oncologic Staging CT OBJECTIVE. The purpose of our study was to determine the prevalence of unsuspected venous thromboembolic disease specifically, of pulmonary embolism (PE) and of inferior vena cava (IVC), iliac, and iliofemoral deep venous thromboses (DVTs) in oncology patients on staging CT scans of the thorax, abdomen, and pelvis. MATERIALS AND METHODS. Retrospective analysis was performed on 435 consecutive staging CT thorax, abdomen, and pelvis scans performed on a variety of oncology patients for the presence of venous thromboembolic disease. The patient group consisted of 198 men and 237 women who ranged in age from 20 to 79 years (mean, 55 years). Disease type, stage, concomitant surgery or chemoradiation therapy regimes, anticoagulation status, and 6-month clinical and radiologic follow-up findings were recorded. RESULTS. We found a prevalence of 6.8% (23/339) unsuspected iliofemoral, 1.2% (4/339) unsuspected common iliac, and 0.3% (1/315) unsuspected IVC DVTs and 3.3% (13/397) unsuspected PEs occurring in patients with a wide range of malignancies. The overall prevalence of unsuspected venous thromboembolism (i.e., DVT, PE, or both) was 6.3% (25/397). DVT, PE, and venous thromboembolic disease were more common in inpatients (p = 0.002, 0.004, 0.023; relative risk [RR] = 1.6, 2.1, 1.4, respectively) and in those with advanced disease (p = 0.001, 0.01, 0.001; RR = 2.2, 1.8, 2.0, respectively). CONCLUSION. Although there is a known increased risk of thromboembolism (DVT and PE) in oncology patients, many cases are not diagnosed, which can prove fatal. Staging CT simultaneously affords one sole investigation of the pulmonary, IVC, iliac, and upper femoral veins, thereby providing an important diagnostic opportunity. Assessment for DVT and PE is important when reviewing staging CT scans. enous thromboembolism has an V incidence of approximately one in 1,000 persons annually [1, 2] and accounts for more than 250,000 hospital admissions annually in the United States [2]. Pulmonary embolism (PE) can be a fatal complication of deep venous thrombosis (DVT) from which more than 100,000 people die annually [3]. More than 90% of PEs arise from lower extremity DVTs [2, 4]. The true prevalence of venous thromboembolism is underestimated because many cases are not apparent clinically. It is estimated that by the time a proximal DVT is diagnosed, PE has already occurred in up to 50% of patients, but only 33 40% of those patients are symptomatic [5, 6]. Malignancy and chemotherapy are independent risk factors and are associated with an increased incidence of venous thrombosis [7]. However, it is estimated that venous thrombosis is diagnosed in fewer than 40% of those with concomitant venous thrombosis and malignancy. Studies have shown that inadequately treated DVT is associated with recurrent PE [8, 9]. When properly diagnosed and treated, clinically apparent PEs are an uncommon cause of death and recur in only a minority of patients [10]. Unsuspected PE can be found in up to 1.5% of routine helical CT scans [11]. The incidence is higher in patients with malignancies, varying from 2.6% to 3.4% [12 14]. To date, the prevalence of unsuspected inferior vena cava (IVC), iliac, and iliofemoral DVT has not been reported. The purpose of our retrospective study was to review staging CT scans for evidence of thromboembolic disease namely, for the presence of IVC; iliac; or iliofemoral DVT, PE, or both. 162 AJR:189, July 2007

CT of Cancer Patients with Asymptomatic Venous Thromboembolic Disease Materials and Methods Twelve examinations were initially excluded because of a known or suspected history of DVT in the prior 12 months. The remaining 435 consecutive staging CT examinations of the thorax, abdomen, and pelvis performed between January 2006 and March 2006 were retrospectively reviewed for the presence of venous thromboembolic disease (198 men, 237 women; age range, 20 79 years; mean age, 55 years). The staging CT examinations were performed for initial staging or restaging. One staging CT examination obtained during the study period was analyzed per patient. These examinations were performed after patients received an IV injection of 150 ml of nonionic contrast medium (240 mg I/mL) administered at a rate of 2 2.5 ml/s. All examinations were performed on a Somatom scanner (Siemens Medical Solutions). CT of the thorax (collimation, 8 mm) was performed 30 seconds after contrast injection. CT of the abdomen and pelvis from the diaphragm to the pubic symphysis (collimation, 8 mm) was performed 90 seconds after the start of contrast medium infusion for CT of the thorax. Disease type, stage, concomitant surgery or chemoradiation therapy regimes, and anticoagulation status were recorded. Cancer stages were divided into two broad groups: early stage disease (stages I and II) and late or advanced stage (stages III and IV). Because this study was retrospective, institutional review board approval was not required. Statistical Analysis SPSS software (version 14, SPSS) was used for statistical analysis. Logistic regression was used to determine the odds ratio (OR) and the statistical association between venous thromboembolism (DVT or PE) and patient age, patient sex, patient category (inpatient or outpatient), disease stage (early or advanced stage), current treatment (receiving or not), and disease type (hematologic, solid organ, or breast). We determined the relative risk (RR) of a thromboembolic event (DVT or PE) in inpatients, those on treatment, and those with advanced stage disease. A p value of 0.05 was considered to indicate a statistically significant difference. Image Interpretation CT scans were independently reviewed by a radiologist and a trainee radiologist. Acquired images were reviewed on a PACS workstation at a window width of 400 H and a window level of 40 H. The reviewers were free to use any window and level settings. Multiplanar reformation images were available on the same workstation. Based on standard diagnostic criteria for diagnosing DVT and PE, filling defects were required to be present on at least two consecutive slices [15]. Exclusion criteria included a known recent (prior 12 months) history of DVT or PE (as determined by radiology records), or a clinical suspicion of DVT or PE at the time of staging scan (as determined by radiology records). We also retrospectively reviewed the medical records of the 25 patients with positive imaging findings for either DVT or PE and 25 randomly chosen patients with negative imaging findings for DVT or PE to assess whether any relevant evidence to suggest a history of DVT and PE had been omitted from the radiology records. Other exclusion criteria included reduced image quality because of excess patient motion, insufficient opacification of the femoral and iliac vessels, and metallic artifacts (caused by metal from hip prostheses), which reduce the diagnostic quality of the scan. Each reviewer recorded confounding variables including lung masses, consolidation, effusions, abdominal masses, and retroperitoneal nodes compressing the abdominopelvic venous system. The location, size, and number of emboli were recorded for each patient. The presence of an embolus and all disagreements were resolved by a consensus panel of three radiologists and one trainee radiologist. There were no discrepancies with regard to the presence of thromboembolism. Follow-Up The radiology reports and clinical notes of the patients with a positive CT examination as judged by the consensus panel were reviewed and followed for 6 months. Follow-up is ongoing. Diagnostic Criteria for DVT and PE The CT criteria for diagnosing PE and IVC, iliofemoral, and iliac DVT using indirect CT venography were similar. A thrombus was defined as a low-attenuating partial or complete intraluminal filling defect surrounded by a high-attenuating ring of enhanced blood and seen on at least two consecutive transverse images. If only one transverse image showed a filling defect, a thrombus was not diagnosed [16, 17]. Ancillary signs of PE, such as infarct or pleural effusions, were not included or used for the PE diagnosis. In our assessment for iliofemoral DVT, a mean venous attenuation (in Hounsfield units) of greater than 90 H was required. This level of contrast is the recommended degree of venous enhancement on CT venography as determined by Bruce et al. [18] and Ghaye et al. [19]. CT can be used to determine whether centrally located thrombi are present within the jugular veins, the brachiocephalic veins, and the superior vena cava and whether an extrinsic process is causing obstruction of the venous channels. However, delayed imaging at 2 3 minutes after contrast injection is required to evaluate these central veins. To our knowledge, no large series has looked at the diagnostic accuracy of this technique in the diagnosis of upper extremity venous thrombosis as other series have in lower extremity venous thrombosis, as outlined previously. Delayed imaging was not performed in our series and therefore precludes assessment for upper limb DVT. Results The groups consisted of 232 patients with early stage disease and 203 with advanced stage disease. Disease types included lymphoma, n = 74; breast carcinoma, n =82; lung cancer, n = 43; gastrointestinal malignancy (esophageal, gastric, small and large bowel, pancreatic cancer, hepatocellular, and biliary), n = 126; urinary tract carcinoma (kidney, bladder, and testicular), n = 44; gynecologic malignancies (ovarian, endometrial, and cervical cancer), n = 34; melanoma, n = 17; and other types of cancer, n = 15. None of the outpatients was on any form of anticoagulation therapy. Scans were obtained before the initiation of chemotherapy in 89 patients, during chemotherapy or an interim treatment in 234 patients, and at follow-up when patients were not undergoing any cancer treatment in 112 patients. Determining the presence of a PE was not confounded by another lung disease in any of the patients. No obstructing or compressing abdominal masses that would contribute to DVT formation were present. Twelve examinations were initially excluded because of a known or suspected DVT within the prior 12 months as determined by reviewing the radiology records. In the remaining 435 examinations that were reviewed, 38 scans (8.7%) were excluded from the evaluation for PE because of poor opacification or excess respiratory motion. The examinations of 96 patients (22%) were excluded from the evaluation for iliofemoral or iliac DVT because of poor opacification or the presence of streak artifact or metallic artifact. Therefore, 397 patients met the inclusion criteria for PE assessment and 339 patients met the inclusion criteria for DVT assessment. Of the 435 patients, 179 were inpatients and 256 were outpatients. After excluding the nondiagnostic scans, 133 of the 339 patients assessed for DVT were inpatients and 206 of the 339 were outpatients; 177 of the 397 patients assessed for PE were inpatients and 220 of the 397 were outpatients (Table 1). Of the 339 patients being evaluated for DVT, we found that 23 (6.8%) had unsuspected iliofemoral DVT and four (1.2%) had AJR:189, July 2007 163

common iliac vein involvement. One of 315 patients (0.3%) evaluated had an IVC DVT, and 13 of 397 patients (3.3%) evaluated had an unsuspected PE (Figs. 1 3). The overall prevalence of unsuspected venous thromboembolism (i.e., DVT, PE, or both) in this group undergoing staging CT of the thorax, abdomen, and pelvis was 6.3% (25/397). Table 2 outlines the demographics (patient sex, patient age, diagnosis, treatment, thromboembolism site) of the 25 patients with positive findings for unsuspected venous thromboembolism. Table 3 outlines the location of the DVTs and PEs and patient characteristics. It highlights the positive venous thrombosis results according to site, number, distribution per disease, inpatient and outpatient status, presence and absence of treatment, and early and advanced stage of disease. All cases of iliofemoral DVT and of iliac DVT showed intraluminal filling defects and two had mural components. None of these patients was on therapeutic anticoagulation at the time of scanning. Fewer than 15% of the A Fig. 1 40-year-old woman (patient 12 in Table 2) with metastatic cervical carcinoma and left iliofemoral deep vein thrombosis (DVT). A and B, CT images show that DVT (white arrows) extends into inferior vena cava in B. Local left iliac and left paraaortic lymph nodes (black arrow, B) are also identified. Other secondary signs of acute DVT on A include venous dilatation compared with normal contralateral side, perivenous soft-tissue infiltration suggestive of edema, contrast material accumulation delineating intraluminal clot, and opacification of collateral veins. TABLE 1: Number and Percentage of Patients Assessed for and with Deep Venous Thrombosis (DVT) and Pulmonary Embolism (PE) CT No. (%) of Patients Finding Inpatients Outpatients Total DVT 14/133 (10.5) 9/206 (4.4) 23/339 (6.8) PE 12/177 (6.8) 1/220 (0.5) 13/397 (3.3) inpatients were taking a prophylactic dose of tinzaparin sodium (Innohep, Pharmion). Patients with advanced disease were more likely to have a DVT or PE, occurring in 8.9% (18/203) of the patients with advanced stage disease; 48% (11/23) of patients with an unsuspected DVT had an unsuspected PE. Table 4 shows the logistic regression analysis results for DVT (chi-square value of 33, degrees of freedom [df] = 6, p = 0.001, n = 339, 2 log likelihood value = 129). Inpatients with a carcinoma were more likely to have a DVT than outpatients (p = 0.002, odds ratio [OR] = 4.8). Those with high-stage disease were significantly more likely to have a DVT than those with a low-stage disease (p = 0.001, OR = 6.3). Patient age, treatment status (receiving or not), and disease type (hematologic, solid organ, or breast) were associated with an increased risk of DVT; however, this finding did not reach statistical significance. Patient sex was not associated with an increased risk of DVT; this finding is not statistically significant. Inpatients or patients with high-stage disease had a higher risk of developing a DVT. According to our results, the relative risk (RR) of DVT in an inpatient with carcinoma is 1.6 (p = 0.044); in a patient with high-stage disease, 2.2 (p = 0.001); and in a patient on treatment, 1.1 (not significant: p = 0.292). Table 5 shows the logistic regression analysis results for PE (chi-square value of 36.3, df =6, p = 0.001, n = 397, 2 log likelihood value = 78). The data in this table show that inpatients with a carcinoma were more likely to have a PE than outpatients (p = 0.004, OR = 22.5). Those with high-stage disease were significantly more likely to have a PE than those with low-stage disease (p = 0.017, OR = 5.8). Patient age was also associated with a risk of PE (p = 0.002, OR = 1). Receiving treatment was associated with an increased risk of PE; however, this finding did not reach statistical significance. Patient sex was not associated with an increased risk of PE; this finding is not statistically significant. Inpatients and patients with high-stage disease had a higher risk of developing a PE. According to our results, the RR of having a PE in an inpatient with carcinoma is 2.1 (p = 0.001); in a patient with high-stage disease, 1.8 (p = 0.019); and in a patient on treatment, 1.2 (not significant: p = 0.56). We further analyzed all the data to assess the overall risk of venous thromboembolism that is, DVT, PE, or both (Table 6). Table 6 shows the logistic regression analysis results for overall venous thromboembolism (chi-square value = 31, df =6, p = 0.001, n = 397, 2 log likelihood value = 149). The data in this table show that inpatients with a carcinoma were more likely to have a venous thromboembolism than outpatients (p = 0.023, OR = 2.8). Those with high-stage disease were significantly more likely to have a venous thromboembolism than those with low-stage disease (p =0.001, OR=5.6). Age was also associated with a risk of venous thromboembolism (p = 0.005, OR = 1). Receiving treatment was associated with an increased risk of venous thromboembo- B 164 AJR:189, July 2007

CT of Cancer Patients with Asymptomatic Venous Thromboembolic Disease A C E Fig. 2 52-year-old man (patient 18 in Table 2) with advanced stage carcinoid. A, CT image shows right iliofemoral deep vein thrombosis (arrow). B and C, CT images show venous thrombosis (arrows) in right iliac vein that extends just distal to confluence of iliac vessels. D and E, CT images show left lower lobar and segmental pulmonary embolus (arrows). B D lism; however, this finding did not reach statistical significance. Patient sex was not associated with an increased risk of venous thromboembolism; this finding is not statistically significant. Therefore, inpatients and patients with high-stage disease have a higher risk of developing a venous thromboembolism. According to our results, the RR of venous thromboembolism in an inpatient with carcinoma is 1.4 (p = 0.047); in a patient with highstage disease, 2 (p = 0.001); and in a patient on treatment, 1.2 (not significant: p = 0.52). At 6-month follow-up, four patients presented with clinically suspected symptomatic PE; three AJR:189, July 2007 165

of these initally had an unsuspected PE and DVT, and one had an unsuspected DVT on staging CT. All returned within 1 2 months of staging CT. Findings on CT pulmonary angiograms were positive in two of the three patients. The recurrent PEs were larger than the initial PE and had more A C E central filling defect on CT pulmonary angiography at the time of the symptomatic PE than at staging CT of unsuspected PE. The third person had negative findings on CT pulmonary angiography, and a Greenfield filter was placed on the basis of the staging CT findings of a DVT and a Fig. 3 57-year-old woman (patient 17 in Table 2) with metastatic esophageal carcinoma. A, CT image shows right iliofemoral deep vein thrombosis (arrow). B E, CT images show lobar and segmental pulmonary emboli (arrows). D and E also show peripheral wedge-shaped infarcts. PE that was now symptomatic. The fourth person with an asymptomatic DVT (no asymptomatic PE) at staging CT had a clinically symptomatic PE. Because there was a strong clinical suspicion for PE in the presence of a known DVT, a Greenfield filter was placed. B D 166 AJR:189, July 2007

CT of Cancer Patients with Asymptomatic Venous Thromboembolic Disease TABLE 2: Patient Diagnosis, Treatment, and Location of Venous Thromboembolism Patient Type and Location of No. Sex Age (y) Diagnosis Treatment Status at CT Venous Thromboembolism Detected 1 F 61 Progressive ovarian cancer Chemotherapy Right iliofemoral DVT 2 F 31 Ovarian cancer After surgery, after chemotherapy Right iliofemoral DVT 3 F 61 Rectal cancer Before surgery, before chemotherapy Right iliofemoral DVT 4 F 65 Unknown primary cancer After chemotherapy Right iliofemoral DVT 5 F 65 Breast cancer Chemotherapy Left iliofemoral DVT 6 M 54 Relapsed lymphoma cancer After chemotherapy Left iliofemoral DVT 7 F 47 Endometrial cancer Chemotherapy Bilateral iliofemoral DVT 8 F 67 Lymphoma, pathologic fracture Chemotherapy Bilateral iliofemoral DVT 9 F 50 Ovarian cancer After chemotherapy Bilateral iliofemoral DVT 10 M 64 Lung cancer After chemotherapy Right iliofemoral DVT, right common iliac DVT 11 F 72 Unknown cancer None Right iliofemoral DVT, right common iliac DVT 12 F 40 Cervical cancer Chemotherapy and radiation therapy Left iliofemoral DVT, IVC DVT 13 F 60 Breast cancer After surgery and before chemotherapy Right iliofemoral DVT, right PE 14 F 54 Advanced metastatic breast cancer Chemotherapy Right iliofemoral DVT, right PE 15 M 59 Lymphoma None Right iliofemoral DVT, left PE 16 F 79 Gastric cancer Before chemotherapy Right iliofemoral DVT, left PE 17 F 57 Esophageal cancer Chemotherapy Right iliofemoral DVT, bilateral PE 18 M 52 Carcinoid After chemotherapy Right iliofemoral DVT, right common iliac DVT, left PE 19 F 50 Breast cancer Chemotherapy Left iliofemoral DVT, left common iliac DVT, left PE 20 F 40 Melanoma Before chemotherapy Left iliofemoral DVT, bilateral PE 21 M 63 Lymphoma a Chemotherapy Left iliofemoral DVT, right PE 22 F 69 Bilateral breast cancer and metastases Chemotherapy Bilateral iliofemoral DVT, right PE 23 F 48 Melanoma Chemotherapy Bilateral iliofemoral DVT, right PE 24 F 53 Breast cancer After surgery and before chemotherapy Bilateral PE 25 M 67 Bladder cancer After surgery Bilateral PE Note DVT = deep venous thrombosis, PE = pulmonary embolism, IVC = inferior vena cava. a Partial response to chemotherapy. Discussion Prompt diagnosis of DVT and PE is essential because both can recur and can have fatal sequelae, thereby affecting morbidity and survival. PE is a fatal complication of DVT from which more than 100,000 people die annually [1]. More than 90% of PEs arise from lower extremity DVT [2]. The cumulated venous thromboembolism recurrence rate is 14% for unilateral proximal DVT and 13.2% for bilateral proximal DVTs. The adjusted survival rate in patients with unilateral proximal DVT is 72% and 65% for those with bilateral proximal DVTs [20]. Patients with cancer and thrombosis have a lower survival rate than those with cancer without thrombosis. Sørensen et al. [21] found that the 1-year survival rate for cancer patients with thrombosis was 12% compared with 36% in control patients (i.e., cancer patients without venous thromboembolism) (p < 0.001). This high mortality was thought to reflect deaths due to both thromboembolism and a more aggressive course of malignancies associated with venous thromboembolism [21]. Patients with cancer have a four- to eightfold higher risk of dying after an acute thrombotic event than patients without cancer [10, 22]. Evidence suggests that the absolute risk of venous thromboembolism depends on the tumor type, the stage or extent of the cancer, and treatment with antineoplastic agents. We confirmed some of these factors in our study. We found a clinically and statistically significant increased prevalence of PE and DVT in inpatients and in those with advanced stage disease. We found a clinical but not statistically significant increased risk of venous thromboembolism in those undergoing treatment. The most common malignancies associated with thrombosis are those of the breast, colon, and lung, reflecting the prevalence of these malignancies in the general population. When data were adjusted for disease prevalence, the cancers most strongly associated with thrombotic complications are those of the pancreas, ovary, and brain [23]. In our patient group, iliofemoral DVT was most frequent in patients with breast cancer, lymphoma, and ovarian cancer. PE was most frequently found in patients with breast cancer, lymphoma, and melanoma. However, on analysis we found no statistical difference in the association of venous thromboembolism with cancer type (hematologic, solid organ, or breast). This may be secondary to the inadequate population size. The diagnosis of venous thromboembolism continues to be a challenge. Venous thromboembolism often has no specific clinical presentation, can be asymptomatic, can masquerade as other illnesses, and may go AJR:189, July 2007 167

TABLE 3: Unsuspected Deep Venous Thrombosis (DVT) and Pulmonary Embolism (PE) Detected on Staging CT Characteristic Location Iliofemoral 18 Iliofemoral and common iliac 4 No. of Patients DVT (n = 23) PE (n =13) Iliofemoral, common iliac, 1 and IVC Unilateral 18 a 4 Bilateral 5 3 Multifocal NA 6 Disease type Breast cancer 6 5 Lymphoma 4 2 Gynecologic cancer 5 0 Melanoma 2 2 Gastrointestinal cancer 3 2 Genitourinary cancer 1 1 Lung cancer 1 0 Other cancer 1 1 Outpatient 9 1 Inpatient 14 12 Disease stage Early (I and II) 5 3 Late (III and IV) 18 10 Treatment status On treatment 10 6 Not on treatment 13 7 Note IVC = inferior vena cava, NA = not applicable. a 12 right, 6 left. undiagnosed. The significance of subsegmental emboli, the role of anticoagulation therapy, and whether the risks of anticoagulation outweigh the benefits in the different PE patient subgroups are other controversial, unsolved areas of study. Advances in radiologic technologies have resulted in improved radiologic diagnosis and identification of incidental DVTs and PEs in asymptomatic patients. Occult PE is a well-recognized clinical entity, and it is believed that most PEs that are fatal are not suspected clinically and are not treated [24, 25]. However, the significance and clinical outcome of patients with asymptomatic DVT and those with asymptomatic PE diagnosed incidentally are not yet known. In previous studies, incidental PEs were detected in 1 1.5% of patients scanned using TABLE 4: Logistic Regression Analysis for Deep Venous Thrombosis 95% Confidence Limits for Exp(B) Variable B SE Wald p Exp(B) Lower Upper Age 0.029 0.015 3.800 0.051 1.030 1.000 1.061 Sex 0.973 0.591 2.709 0.100 0.378 0.119 1.204 Inpatient vs outpatient 1.579 0.504 9.832 0.002 4.852 1.808 13.023 Disease stage 1.843 0.569 10.473 0.001 6.313 2.068 19.269 Treatment status 0.513 0.488 1.109 0.292 1.671 0.643 4.345 Disease type 1 0.270 0.694 0.151 0.697 1.310 0.336 5.099 2 0.137 0.589 0.054 0.816 1.147 0.361 3.642 Note Variables entered on step 1: age, sex, in- versus outpatient, disease stage, treatment status, disease type. B = coefficient (B), Exp(B) = Wald statistic or estimated odds ratio. TABLE 5: Logistic Regression Analysis for Pulmonary Embolism 95% Confidence Limits for Exp(B) Variable B SE Wald p Exp(B) Lower Upper Age 0.068 0.022 9.920 0.002 1.070 1.026 1.116 Sex 0.697 0.709 0.968 0.325 0.498 0.124 1.998 Inpatient vs outpatient 3.114 1.073 8.424 0.004 22.510 2.749 184.335 Disease stage 1.762 0.739 5.690 0.017 5.826 1.369 24.790 Treatment status 0.581 0.634 0.839 0.360 1.787 0.516 6.190 Disease type 1 0.108 0.726 0.022 0.882 1.114 0.269 4.624 2 0.880 0.683 1.657 0.198 0.415 0.109 1.584 Note Variables entered on step 1: age, sex, in- versus outpatient, disease stage, treatment status, disease type. B = coefficient (B), Exp(B) = Wald statistic or estimated odds ratio. TABLE 6: Logistic Regression Analysis for Overall Venous Thromboembolism (Deep Venous Thrombosis, Pulmonary Embolism, or Both) 95% Confidence Limits for Exp(B) Variable B SE Wald p Exp(B) Lower Upper Age 0.040 0.014 7.717 0.005 1.041 1.012 1.070 Sex 0.828 0.532 2.427 0.119 0.437 0.154 1.238 Inpatient vs outpatient 1.052 0.461 5.200 0.023 2.864 1.159 7.074 Disease stage 1.729 0.542 10.192 0.001 5.635 1.949 16.288 Treatment status 0.387 0.458 0.715 0.398 1.473 0.600 3.616 Disease type 1 0.306 0.335 0.830 0.362 1.358 0.703 2.620 2 0.579 0.556 0.589 0.443 0.415 0.678 1.989 Note Variables entered on step 1: age, sex, in- versus outpatient, disease stage, treatment status, disease type. B = coefficient (B), Exp(B) = Wald statistic or estimated odds ratio. single-detector CT and IV contrast material [11, 26]. Both of these studies found a higher rate of PEs in a subgroup of cancer patients than in those with other risk factors such as trauma, surgery, congestive cardiac failure, and immobilization. Gosselin et al. [11] and Winston et al. [26] also found a higher rate of PEs among inpatients of between 2% and 5%. From their subgroup analyses, Winston et al. found a PE prevalence of 9% in inpatients 168 AJR:189, July 2007

CT of Cancer Patients with Asymptomatic Venous Thromboembolic Disease with malignancies. Sebastian and Paddon [14] found a prevalence of 2.6% in oncology patients, but they did not perform subgroup analysis by cancer type or cancer stage or by whether patients were inpatients or outpatients. Gladish et al. [12] found a PE prevalence of 4% in oncology patients (6% in oncology inpatients). Storto et al. [13] found a prevalence of 3.4 4% in inpatients with malignancy and 0.9% in outpatients with malignancy. These results are similar to our finding of an overall prevalence of 3.3% in outpatients and 6.8% in inpatients. To our knowledge, no long-term survival studies of patients with an incidentally detected DVT or PE have been reported in the literature. Engelke et al. [27] in a recent retrospective study found that those with an incidentally diagnosed and undetected PE at the time of scanning, despite failure to treat, had a benign prognosis. Their results concurred with those of Schultz et al. [28] in their review of untreated PEs in trauma patients. However, the PEs in both of these studies were of low intrapulmonary clot burden (Engelke et al.) or were minor PEs (Schultz et al.). These retrospective trials were also limited by small patient numbers, and both groups of investigators recommended further assessment. Low intrapulmonary clot burden and minor PEs are also in keeping with the current dilemma regarding the clinical importance and risk versus benefit of anticoagulation therapy for treatment of small subsegmental emboli discovered at MDCT [29 31]. In our group, we found significant clot burden and in each case the clot was located in the lobar or segmental branches. The prevalence of IVC, iliac, and iliofemoral DVTs has not been reported to date. We found an overall incidental finding of iliofemoral DVT in 6.8% of patients. Iliofemoral DVT was found in 10% of the inpatient group, most of whom had advanced stage malignancy. An incidental finding of DVT in this patient group is important because 8.7% (2/23) with an asymptomatic DVT and PE subsequently had a more significant symptomatic PE diagnosed on CT pulmonary angiography within 8 weeks. Another 8.7% (2/23) of patients with an asymptomatic DVT (one of whom also had an asymptomatic PE) had clinically symptomatic PEs requiring Greenfield filter insertion within 6 weeks of imaging. Limitations A limitation of staging CT as a means of venous thromboembolism assessment versus CT pulmonary angiography is collimation size. In 1992, Remy-Jardin et al. [32] used 5- or 10-mm helical CT, which is similar to the collimation that we used. They found a high sensitivity and specificity (80 100%) for detection of central PEs. This scanning method was noted to be less sensitive for the detection of segmental and subsegmental PEs than other diagnostic techniques [32, 33]. Therefore, subsegmental PEs may be missed on staging CT. However, currently a smaller collimation size in staging CT is now standard practice, which would almost certainly increase the sensitivity and specificity for the detection of subsegmental PEs. Gladish et al. [12] used a collimation of 3.75 mm. However, as we stated earlier, the appropriate management of subsegmental PEs incidentally detected is currently controversial. Eyer et al. [34] in a retrospective review of clinicians responses to radiologists reports of isolated subsegmental PE using MDCT found that the patients received anticoagulation therapy more often than not. In those patients who did not receive anticoagulation therapy for subsegmental PEs, no recurrent PE was identified on follow-up [34]. Other limitations of our study besides collimation size include the retrospective method of data collection and the lack of upper extremity venous evaluation. Excluding nondiagnostic scans due to inadequate vessel opacification may lead to the failure to diagnose a DVT or PE. Nevertheless, these findings are significant and important. Attention should be paid to these areas on review of staging CT. Conclusion We have shown that cancer patients have a significantly increased rate of incidental venous thromboembolism (iliofemoral DVT and PE) on staging CT scans than the general population. Thromboembolism was seen more frequently in inpatients (RR of DVT, PE, venous thromboembolic disease = 1.6, 2.1, 1.4, respectively) and in those with advanced stage disease (RR of DVT, PE, venous thromboembolic disease = 2.2, 1.8, 2.0). Therefore, a high index of suspicion and vigilance and dedicated assessment for thromboembolism are recommended while interpreting all scans of oncologic patients regardless of cancer stage. Staging CT of the thorax, abdomen, and pelvis provides a single examination that allows both the pulmonary arterial system and the pelvic and lower extremity venous system to be evaluated, therefore offering an important diagnostic opportunity. References 1. White RH. The epidemiology of venous thromboembolism. Circulation 2003; 107:1 4 2. Haines ST. Venous thromboembolism: pathophysiology and clinical presentation. Am J Health Syst Pharm 2003; 60[22 suppl 7]:S3 S5 3. Goldhaber SZ. Pulmonary embolism. N Engl J Med 1998; 339:93 104 4. Silverstein MD, Heit JA, Mohr DN, Petterson TM, O Fallon WM, Melton LJ III. Trends in the incidence of deep vein thrombosis and pulmonary embolism. Arch Intern Med 1998; 158:585 593 5. Huisman MV, Buller HR, ten Cate JW, et al. Unexpected high prevalence of silent pulmonary embolism in patients with deep venous thrombosis. Chest 1989; 95:498 502 6. Dorfman GS, Cronan JJ, Tupper TB, Messersmith RN, Denny DF, Lee CH. Occult pulmonary embolism: common occurrence in deep venous thrombosis. AJR 1987; 148:263 266 7. Sørensen HT, Mellemkjær L, Flemming, HO, Nielsen GL. The risk of a diagnosis of cancer after primary deep venous thrombosis or pulmonary embolism. N Engl J Med 1998; 17 338:1169 1173 8. Hull R, Delmore T, Carter C, et al. Adjusted subcutaneous heparin versus warfarin sodium in the longterm treatment of venous thrombosis. N Engl J Med 1982; 306:189 194 9. Hull R, Hirsh J, Jay R, et al. Different intensities of oral anticoagulant therapy in the treatment of proximal-vein thrombosis. N Engl J Med 1982; 307:1676 1681 10. Carson JL, Kelley MA, Duff A, et al. The clinical course of pulmonary embolism. N Engl J Med 1992; 19:326:1240 1245 11. Gosselin MV, Rubin GD, Leung AN, Huang J, Riczk NW. Unsuspected pulmonary embolism: prospective detection on routine helical CT scans. Radiology 1998; 208:209 215 12. Gladish GW, Choe DH, Marom EM, Sabloff BS, Broemeling LD, Munden RF. Incidental pulmonary emboli in oncology patients: prevalence, CT evaluation, and natural history. Radiology 2006; 240:246 255 13. Storto ML, Di Credico A, Guido F, Larici AR, Bonomo L. Incidental detection of pulmonary emboli on routine MDCT of the chest. AJR 2005; 184:264 267 14. Sebastian AJ, Paddon AJ. Clinically unsuspected pulmonary embolism: important secondary finding in oncology CT. Clin Radiol 2006; 61:81 85 15. Brunot S, Corneloup O, Latrebe V, Montaudon M, Laurent F. Reproducibility of multidetector spiral computed tomography in detection of subsegmental acute pulmonary embolism. Eur Radiol 2005; 15:2057 2063 16. Remy-Jardin M, Remy J, Deschildre F, et al. Diagnosis of pulmonary embolism with spiral CT: comparison with pulmonary angiography and scintigraphy. Radiology 1996; 200:699 706 AJR:189, July 2007 169

17. Tillie-Leblond I, Masrota I, Radenne F, et al. Risk of pulmonary embolism after a negative spiral CT angiogram in patients with pulmonary disease: 1- year clinical follow-up study. Radiology 2002; 223:461 467 18. Bruce D, Loud PA, Klippenstein DL, Grossman ZD, Katz DS. Combined CT venography and pulmonary angiography: how much venous enhancement is routinely obtained? AJR 2001; 176:1281 1285 19. Ghaye B, Szapiro D, Willems V, Dondelinger RF. Pitfalls in CT venography of lower limbs and abdominal veins. AJR 2002; 178:1465 1471 20. Seinturier C, Bosson JL, Colonna M, Imbert B, Carpentier PH. Site and clinical outcome of deep vein thrombosis of the lower limbs: an epidemiological study. J Thromb Haemost 2005; 3:1362 1367 21. Sørensen HT, Mellemkjaer L, Olsen JH, Baron JA. Prognosis of cancers associated with venous thromboembolism. N Engl J Med 2000; 343:1846 1850 22. Prandoni P, Lensing AW, Cogo A, et al. The longterm clinical course of acute deep venous thrombosis. Ann Intern Med 1996; 125:1 7 23. Lee AYY, Levine ML. Venous thromboembolism and cancer: risks and outcomes. Circulation 2003; 107:1 17 24. Ryu JH, Olson EJ, Pellikka PA. Clinical recognition of pulmonary embolism: problem of unrecognized and asymptomatic cases. Mayo Clin Proc 1998; 73:873 879 25. Pineda LA, Hathwar VS, Grant BJ. Clinical suspicion of fatal pulmonary embolism. Chest 2001; 120:791 795 26. Winston CB, Wechsler RJ, Salazar AM, Kurtz AB, Spirn PW. Incidental pulmonary emboli detected at helical CT: effect on patient care. Radiology 1996; 201:23 27 27. Engelke C, Rummeny EJ, Marten K. Pulmonary embolism at multi-detector row CT of chest: oneyear survival of treated and untreated patients. Radiology 2006; 239:563 575 28. Schultz DJ, Brasel KJ, Washington L, et al. Incidence of asymptomatic pulmonary embolism in moderately to severely injured patients. J Trauma 2004; 56:727 731 29. Ghaye B, Remy J, Remy-Jardin M. Non-traumatic thoracic emergencies: CT diagnosis of acute embolism the first 10 years. Eur Radiol 2002; 12:1886 1905 30. Dalen JE. Pulmonary embolism: what have we learned since Virchow? Natural history, pathophysiology, and diagnosis. Chest 2002; 122:1440 1456 31. Egermayer P, Town GI. The clinical significance of pulmonary embolism: uncertainties and implications for treatment a debate. J Intern Med 1997; 214:5 10 [Erratum in J Intern Med 1997; 214:341] 32. Remy-Jardin M, Remy J, Wattine L, Giraud F. Central pulmonary thromboembolism: diagnosis with spiral volumetric CT with the single-breath-hold technique comparison with pulmonary angiography. Radiology 1992; 185:381 387 33. Blum AG, Delfau F, Grignon B, et al. Spiral computed tomography versus pulmonary angiography in the diagnosis of acute massive pulmonary embolism. Am J Cardiol 1994; 74:96 98 34. Eyer BA, Goodman LR, Washington L. Clinicians response to radiologists reports of isolated subsegmental pulmonary embolism or inconclusive interpretation of pulmonary embolism using MDCT. AJR 2005; 184:623 628 170 AJR:189, July 2007