High rates of symptomatic and incidental thromboembolic events in gastrointestinal cancer patients


Alok A. Khorana, 601 Elmwood Ave, Box 704, Rochester, NY 14642, USA.
Tel.: +1 585 275 4797; fax: +1 585 273 1042.
E-mail: alok_khorana@URMC.rochester.edu

The association of thromboembolism with cancer, although known for over a century, has gained increased attention in recent years. Thromboembolic complications include deep vein thrombosis (DVT), pulmonary embolism (PE) and arterial events including stroke and myocardial infarction. Cancer patients on active treatment are particularly at risk, with a recent study demonstrating a 47% increase in frequency in hospitalized cancer patients receiving chemotherapy [1]. There are significant consequences of venous thromboembolism (VTE) in the cancer patient, including an association with mortality, a high risk of recurrent VTE and a paradoxically high risk of bleeding complications [2–4]. Patients with gastrointestinal cancers in particular have amongst the highest rates of VTE [5,6].

One of the potential causes of the increased frequency of VTE is the improvement in the resolution of computed tomography (CT) technology in recent years. In the United States, multidetector-row CT technology expanded considerably in the earlier part of this decade [7]. This has led to an increased frequency of incidentally detected thromboembolic events. Recent analyzes have explored the implications of such incidental events and it has been questioned whether these are truly asymptomatic events [8]. Additionally, in a recent retrospective analysis, 6-month mortality was similar in cancer patients with symptomatic or asymptomatic VTE but significantly higher in comparison to cancer patients without VTE [9]. These analyzes have focused primarily on incidental events. Little is known about the prevalence and relative proportion of symptomatic and incidental thromboembolic events.

The objective of the present study was to determine the prevalence and relative proportions of incidental and symptomatic thromboembolic events in high-risk cancer patients. We conducted a retrospective analysis of a cohort of consecutive patients with gastrointestinal cancers undergoing active treatment at the Gastrointestinal Cancer Program of the James P. Wilmot Cancer Center at the University of Rochester Medical Center from 1 July 2008 to 31 December 2008. Primary sites of cancer considered for inclusion were esophageal, gastric including gastro-esophageal junction, liver, biliary tract, pancreatic, small bowel, colon, rectum and anal. Histologic subtypes included adenocarcinomas, squamous cell carcinomas and neuroendocrine tumors (carcinoids). Two hundred and thirty consecutive patients were initially identified through a clinical database of active patients maintained by the Gastrointestinal Cancer Program. Ten patients were excluded owing to lack of availability of scan results. Clinicopathologic information on these patients was obtained primarily through chart review and examination of CT scan and ultrasonography scan results. All incidentally discovered VTE were diagnosed on routine restaging scans; screening for DVT/PE was not part of institutional policy. Data gathered included site and stage of cancer, patient characteristics, site of DVT, PE, arterial events or other thrombotic events. Charts were also reviewed to evaluate for symptoms suggestive of DVT or PE including cough, hemoptysis, chest pain, abdominal pain, new ascites and extremity swelling or tenderness.

Institutional review board approval was obtained to conduct this retrospective study. Patient confidentiality was maintained with the patient database kept in a data file with access limited to study authors.

A total of 220 consecutive gastrointestinal cancer patients were included in the study. Of these, 126 (57.3%) were men and 94 (42.7%) were women. The mean age of the study population was 61 years (range, 24 to 91 years). Common primary sites of cancer included the colon (= 57, 25.9%), pancreas (= 48, 21.8%), esophagus (= 39, 17.7%), liver (= 33, 15%), stomach (= 13, 5.9%), rectum (= 12, 5.5%) and gall bladder (= 7, 3.2%), among others. The majority of patients had metastatic disease (= 154, 70%).

Sixty patients (27.3%) were found to have experienced a total of 83 thromboembolic events (Table 1). These included 32 DVTs (38.6%) and 17 PEs (20.5%). An additional 23 patients developed 25(30.1%) visceral vein or 9 (10.8%) arterial thromboembolic events. Six patients had catheter-related DVTs and an additional patient had a DVT at the site of an unsuccessful attempt at catheter placement. Ten of the 60 patients experienced more than one event, including seven patients with both DVT and PE, three with DVT and other thrombi (two with portal and one with superior mesenteric vein) and two patients with DVT, PE and other thrombi (one with thrombi in portal, splenic, superior mesenteric vein and superior mesenteric artery and the other with an aortic thrombus).

Table 1.   Sites of thromboembolic events
Type n (%)
All83 (100)
DVT32 (38.6%)
Lower extremity veins
Upper extremity veins
Neck veins
 Superior vena cava3
 Inferior vena cava4
 Internal jugular6
 External jugular1
PE17 (20.5%)
Visceral vein25 (30.1%)
 Superior mesenteric6
 Inferior mesenteric1
Arterial9 (10.8%)
 Superior mesenteric1

Out of 32 patients with DVTs, 16 (50%) were incidentally discovered. Of the 17 patients who had PE, 6 (35.3%) were asymptomatic for shortness of breath, fatigue, cough, chest pain or hemoptysis. An additional six patients reported symptoms of shortness of breath (four of six aroused clinical suspicion for VTE), five reported fatigue (one aroused suspicion) and one reported a history of cough (but did not arouse suspicion for VTE).

There were 34 events in an additional 23 patients with other sites of thrombotic involvement, including six with more than one blood vessel involvement. Common sites of visceral vein involvement included portal, superior mesenteric and splenic veins.

Thus, in a retrospective study of gastrointestinal cancer patients receiving active chemotherapy at our institution we found a high prevalence of thrombotic events, with over one-fourth of all patients experiencing at least one event. DVTs, as expected, were the most common followed by visceral vein thromboses, PEs and arterial events. A novel finding of this study was the observation that a high proportion of these events were asymptomatic, including 50% of DVTs and over 35% of PEs.

The discovery of incidental PE on routine restaging scans conducted in cancer patients has increasingly become a concern with the advent of multidetector row CT technology [7,10]. Varying prevalence rates have been noted in recent studies. In a CT evaluation study of 403 cancer patients in whom adequate quality multidetector row thoracic CT was performed for indications other than PE assessment, 16 patients (4%) had PE, although the initial report identified only four of them [11]. In a retrospective study of 435 staging CT scans performed in cancer patients, a prevalence rate of 6.3% for unsuspected VTE was reported [12]. This included rates of 6.8% for iliofemoral DVT, 1.2% for common iliac, 0.3% for IVC DVTs and 3.3% for PE. In our study, the rate of unsuspected DVT was 7.3% and the rate of unsuspected PE was 2.7%, somewhat higher than these prior reports. However, our study population was comprised specifically of gastrointestinal cancer patients receiving chemotherapy, 70% of whom had metastatic disease. All of these are significant risk factors for VTE [13].

Another notable finding of our study was the high prevalence of visceral vein thrombi. In our analysis, we found nearly one-third of venous events occurred in the visceral veins. Portal vein thrombi were the most common, followed by superior mesenteric and splenic vein thrombi. Portal vein thrombosis has previously been reported in gastrointestinal cancers [14–16]. In hepatocellular carcinoma, it has been associated with both systemic VTE and poor survival suggesting a systemic hypercoagulable state rather than simply a local thrombus [14,15]. Little is known about the natural history of visceral vein thrombi and need for anticoagulation. At our institution, the most common practice is to not anticoagulate asymptomatic patients with visceral vein thrombi.

Limitations of this study included its retrospective nature and single-institution design. Symptoms experienced by patients but either not reported to or not recorded by providers would not be identified on chart review; to ascertain the true prevalence of asymptomatic VTE a prospective cohort study or registry is necessary. A single-institution study does, however, limit variation in practice patterns for diagnosis of VTE. Rates may be underestimated as this study was designed as a prevalence rather than incidence study and many patients in this population continue to receive active treatment. As lower extremity DVT involving the common femoral, popliteal and other more distal veins would not be seen on routine restaging scans, the rates of asymptomatic VTE in this study are likely to have been further underestimated.

In summary, venous and arterial thromboembolic events are highly frequent in gastrointestinal cancer patients receiving systemic chemotherapy. A large proportion of these events are unsuspected and identified on routine restaging scans, including one-half of all DVTs and over one-third of PEs. Asymptomatic visceral vein thrombi are also common. Further research is necessary to investigate the natural history of asymptomatic thrombi, need for anticoagulation and association with subsequent systemic VTE and survival.


A.A. Khorana is supported by grants from the National Cancer Institute K23 CA120587, the National Heart, Lung and Blood Institute 1R01HL095109-01 and the V Foundation. The authors would like to acknowledge J. Berkhof, C. Copoulos and K. DeGrave for their contributions to the clinical database.

Disclosure of Conflict of Interests

The authors state that they have no conflict of interest.