Comparison between idiopathic deep vein thrombosis of the upper and lower extremity regarding risk factors and recurrence


Paul A. Kyrle, Department of Medicine I, Medical University of Vienna, Waehringer Guertel 18-20, A-1090 Vienna, Austria.
Tel.: +43 1 40400 4951; fax: +43 1 40400 4030.


Summary. Background: The pathogenesis and natural course of idiopathic upper extremity deep vein thrombosis (UEDVT) are unclear. Objective: To compare patients with UEDVT and with idiopathic lower extremity deep vein thrombosis (LEDVT) regarding risk factors and recurrence. Methods: We followed 50 patients with first idiopathic UEDVT and 841 patients with first idiopathic LEDVT for an average of 59 and 46 months, respectively. We excluded patients with natural inhibitor deficiency, lupus anticoagulant, cancer, pregnancy, isolated pulmonary embolism (PE), or long-term antithrombotic treatment. The endpoint was recurrent venous thromboembolism (VTE). Results: In comparison to LEDVT patients, UEDVT patients were younger (38 ± 13 years vs. 49 ± 16 years, P < 0.001), slimmer (body mass index: 24 ± 4 vs. 27 ± 5, P < 0.001), less frequently had a family history of VTE (18% vs. 31%, P = 0.06) or concomitant PE (8% vs. 31%, P =0.001), were less frequently carriers of factor V Leiden (12% vs. 30%, P = 0.009), and had lower thrombin generation marker levels (D-dimer, 283 ± 361 ng mL−1 vs. 456 ± 446 ng mL−1, P < 0.001; peak thrombin, 298 ± 101 nm vs. 363 ± 111 nm, P = 0.001). Recurrence occurred in two of 50 patients with UEDVT (4%) and in 129 of 841 patients with LEDVT (15%). After 5 years, the likelihood of recurrence was 2% [95% confidence interval (CI) 0–6] among UEDVT patients and 19% (95% CI  16–22; P = 0.02) among LEDVT patients. As compared to LEDVT patients, the adjusted risk of recurrence was 0.26 (95% CI  0.06–1.05; P = 0.059) in UEDVT patients. Conclusion: The pathogenesis and natural course of the disease differ between patients with idiopathic UEDVT and LEDVT.


Upper extremity deep vein thrombosis (UEDVT) accounts for up to 14% of deep vein thrombosis (DVT) cases [1,2] and is often linked to temporary risk conditions such as insertion of a central venous catheter, malignancy, arm surgery or trauma, or immobilization by plaster cast [1,3]. In contrast to the situation in patients with lower extremity deep vein thrombosis, the incidence of laboratory thrombophilia, in particular of factor V (FV) Leiden, appears to be low among patients with UEDVT. Even so, Martinelli et al. reported odds ratios for UEDVT patients of 6.2 for FV Leiden, 5.0 for prothrombin G20210A, and 4.9 for deficiency of a natural coagulation inhibitor [4]. Idiopathic UEDVT, that is, DVT in the absence of a triggering temporary risk factor, is called the Paget–Schroetter syndrome and can occur after physically strenuous use of the arm and shoulder or can result from venous compression at the thoracic outlet. Complications of UEDVT are pulmonary embolism (PE) (reported in up to 36% of patients), the post-thrombotic syndrome, and recurrence [4–6].

Within the frame of the Austrian Study on Recurrent Venous Thromboembolism (AUREC), a large ongoing prospective cohort study, we aimed to study the natural course of idiopathic UEDVT. We were particularly interested in differences between patients with UEDVT and LEDVT regarding the frequency of clinical and laboratory risk factors and the risk of recurrent venous thromboembolism (VTE).

Materials and methods

Patients and study design

This study was performed within the frame of AUREC, a large prospective ongoing cohort study in patients with VTE. Detailed information on AUREC has been provided [7]. Briefly, patients were included when they were older than 18 years, had a symptomatic VTE, which was confirmed by objective investigations, and were treated with anticoagulants for at least 3 months. Exclusion criteria were: a previous episode of VTE; VTE related to surgery, trauma, or pregnancy; deficiency of antithrombin, protein C, or protein S; presence of the lupus anticoagulant; cancer; or requirement for long-term antithrombotic treatment for reasons other than VTE. Patients entered the study at the time of discontinuation of oral anticoagulation. At study entry, a detailed history was taken and a detailed physical examination was performed. A family history of VTE was defined if one or more first-degree family member(s) had a venous thromboembolic event. Measurement of antithrombin, protein C, protein S, the lupus anticoagulant, FV Leiden, prothrombin G20210A, factor VIII (FVIII), D-dimer and in vitro thrombin generation (peak thrombin) was performed at least 3 weeks after withdrawal of anticoagulation. Patients were observed at 3-month intervals for the first year and every 6 months thereafter. They were given detailed written information on symptoms of VTE and were asked to report immediately if such symptoms occurred. A medical history was obtained at each visit. Female patients were strongly discouraged from intake of oral contraceptives or hormone replacement therapy. Patients received routine thromboprophylaxis with a low molecular weight heparin during high-risk situations such as prolonged immobilization or long-haul air travel. The ethics committee of the Medical University of Vienna approved the study. Informed consent was obtained from all participants.

Diagnosis of venous thromboembolism

The diagnosis of VTE was established by a positive finding on venography or color duplex sonography (in the case of proximal DVT). If venography was applied, at least one of the following direct or indirect criteria had to be fulfilled: a constant filling defect seen on two views; an abrupt discontinuation of the contrast-filled vessel at a constant level of the vein; and the entire deep vein system without external compression failing to fill, with or without venous flow through collateral veins. Diagnostic criteria for color duplex ultrasonography were as follows: visualization of an intraluminal thrombus in a deep vein, incomplete or absent compressibility, and lack of flow spontaneously and after distal manipulation.

A diagnosis of PE was considered in cases of typical symptoms, that is, chest pain, dyspnea, cough, hemoptysis, and/or syncope. The diagnosis of PE was confirmed either by ventilation/perfusion lung scanning according to the criteria of the Prospective Investigation of Pulmonary Embolism Diagnosis [8] or by a spiral computed tomography (CT) scan demonstrating one or several low-attenuation areas that partly or completely filled the lumen of an opacified vessel.

Definition of anatomic compression and unusual activity

A systematic diagnostic work-up for evaluation of an anatomic compression was not performed. The diagnosis of an anatomic compression was based on radiologic findings obtained by CT scan or magnetic resonance imaging (MRI) and included musculoskeletal abnormalities and narrowness of the interscalene space, but did not include a uniform definition of an anatomic compression. For diagnosis of an anatomic compression due to a cervical rib, chest X-ray was permitted.

Strenuous exercise of the affected arm was defined either as single exceptional activities carried out within a few days before onset of DVT symptoms or as strenuous activities that were started a few weeks earlier and that were repeatedly performed until the onset of DVT symptoms, for example, in the course of sport activities.

Study endpoints

The endpoint of the study was recurrent symptomatic DVT confirmed by venography or color duplex sonography, or recurrent symptomatic PE confirmed by ventilation/perfusion lung scanning and/or spiral CT scan according to the aforementioned criteria. Recurrent DVT was diagnosed if the patient had a thrombus in another deep vein in the extremity involved in the previous event, a thrombus in the opposite extremity, or a thrombus in the same venous system with a proximal extension of the thrombus (if the upper limit of the original thrombus had been visible) or the presence of a constant filling defect surrounded by contrast medium (if the original thrombus had not been visible).

Laboratory analysis

Venous blood from fasting patients was collected into 1 : 10 volume of 0.11 mm trisodium citrate and immediately centrifuged for 20 min at 2000 × g. Aliquots of plasma were stored at − 80 °C until analysis. Genomic DNA was isolated from leukocytes by standard methods. Technicians were unaware of patients’ characteristics at all times.

Screening for FV Leiden and prothrombin G20210A and measurement of antithrombin, protein C, protein S, homocysteine, FVIII and D-dimer were carried out as previously described [9–12]. The diagnosis of the lupus anticoagulant was based on criteria of the International Society on Thrombosis and Haemostasis [13]. Peak thrombin generation was determined using a fluorescence-based immunoassay kit (Technothrombin TGA; Technoclone, Vienna, Austria) [14].

Statistical analysis

Categorical data were compared between groups using contingency-table analyses (the chi-square test). Continuous data (presented as means ± SD) were compared with use of Mann–Whitney U-tests. All P-values were two-tailed. Survival-time methods were used to analyze the time to recurrence among patients with a subsequent episode of VTE (uncensored observation) or the duration of follow-up among patients without recurrence (censored observations) [15]. The probability of recurrence was estimated according to the method of Kaplan and Meier [16]. Data on patients who left the study or who were lost to follow-up were censored at the time of withdrawal. To test for homogeneity between strata, we applied the log-rank test. Univariate and multivariate Cox proportional-hazards models were used to analyze the association between the locations of the index DVT (upper or lower extremity) and the risk of recurrent VTE. Analyses were adjusted for age, sex, the presence or absence of FV Leiden or prothrombin G20210A, hyperhomocysteinemia (dichotomized at the 95th percentile of the reference range), and FVIII (dichotomized at a plasma level of 230 IU dL−1). The cut-off values for continuous data were chosen on the basis of previous analyses showing the best discrimination between low-risk and high-risk patients [10,17]. All computations were performed with the use of SPSS software, version 15.0.


Study population

Between July 1992 and October 2006, 3322 VTE patients older than 18 years were screened. Of these, 2165 were excluded because of the following conditions: long-term oral anticoagulation for reasons other than VTE (488 patients); VTE associated with surgery, trauma, or pregnancy (503); cancer (480); more than one episode of VTE (440); presence of the lupus anticoagulant (80); and antithrombin, protein C or protein S deficiency (82). Twenty-four patients with more than one thrombophilic condition were excluded because they received long-term anticoagulation. We excluded 31 patients with FVIII levels greater than 230 IU dL−1 who were included in a trial investigating the effect of long-term anticoagulant therapy or received long-term anticoagulation. Thirty-seven patients were excluded because they participated in other trials. For the present analysis, we further excluded 266 patients with symptomatic PE without clinical symptoms suggestive of LEDVT or UEDVT.

The study population therefore consisted of 891 patients, of whom 50 (5.6%) had idiopathic UEDVT and 841 (94.4%) had idiopathic LEDVT. A total of 242 patients left the study because they required antithrombotic treatment for reasons other than VTE (139 patients), were diagnosed with cancer during follow-up (21 patients), became pregnant and were started on low molecular weight heparin prophylaxis (40 patients), or were lost to follow-up (29 patients). Thirteen patients died during follow-up: cardiac failure (six), cancer (one), sepsis (one), choking asphyxia (one), cerebral infarction (one), cerebral hemorrhage (one), aortic dissection (one), and suicide (one). Data on these patients were censored at the time of death.

Patient characteristics

As compared with patients with LEDVT, those with UEDVT were on average younger and slimmer, and had more right-sided events (Table 1). Importantly, they also suffered less frequently from concomitant PE (8% vs. 31%, P = 0.001). There was no difference between UEDVT and LEDVT patients with respect to sex distribution. Among women aged 50 years or less, oral contraceptive use was more frequent among patients with LEDVT. Patients with a UEDVT less frequently had a family history of VTE. The mean duration of anticoagulation was shorter among patients with UEDVT than among those with LEDVT, whereas follow-up times after discontinuation of anticoagulant treatment did not differ significantly.

Table 1.   Clinical characteristics, risk factors and hemostatic system activation of patients with upper extermity deep vein thrombosis (UEDVT) and lower extremity deep vein thrombosis (LEDVT)
 UEDVT (n = 50)LEDVT (n = 841)P
  1. VTE, venous thromboembolism; PE, pulmonary embolism; cut-off value for high factor VIII: 230 IU dL−1.

Male gender (%)23 (46)406 (48)0.75
Age (years)38 ± 1349 ± 16< 0.001
Body mass index (kg m−2)24 ± 427 ± 5< 0.001
Family history of VTE (%)9 (18)256 (31)0.06
Oral contraceptive use (%) 10/20 (50)190/262 (73)0.03
Thrombosis right-sided (%)31 (62)365 (44)0.01
Concomitant PE (%)4 (8)258 (31)0.001
Anticoagulation (months)6.7 ± 4.27.9 ± 8.70.03
Follow-up (months)59 ± 4746 ± 400.12
Factor V Leiden (%)6 (12)246 (30)0.009
Prothrombin G20210A (%)3 (6)53 (6)0.99
High factor VIII level (%)3 (6)72 (9)0.55
Hyperhomocysteinemia (%)12 (27)201 (26)0.83
D-dimer (ng mL−1)283 ± 361456 ± 446< 0.001
Peak thrombin (nm)298 ± 101363 ± 1110.001

Among LEDVT patients, thrombosis was confined to the calf veins in 257 patients (31%), whereas 584 patients (69%) had DVT. UEDVT involved the axillary and/or subclavian vein in 36 patients (72%), the axillary and/or subclavian vein and the brachial vein in 12 patients (24%), and the brachial vein in only two patients (4%). A CT scan or MRI was performed in 27 UEDVT patients. In two patients, an anatomic compression was diagnosed. One patient had a narrowed distance between the posterior border of the clavicle and the superior margin of the first rib, and the other had muscular hypertrophy. Chest X-ray, but neither MRI nor a CT scan, was performed in 22 patients. In two patients, a cervical rib was diagnosed. One patient with a cervical rib underwent surgical decompression after cessation of secondary thromboprophylaxis. None of the patients with persistent anatomic compression had recurrent VTE during follow-up periods of 22, 55, 82 and 115 months, respectively.

Ten patients (20%) reported strenuous use of the affected arm (weight-lifting in seven patients and unusual sport activities with prolonged repeated abduction in three patients) shortly before DVT occurrence. Information about the dominant arm was available for 36 patients (33 patients were right-handed). In 22 patients (61%), UEDVT took place in the dominant arm.

Thrombophilic risk factors and coagulation activation markers

The proportion of carriers of FV Leiden was significantly lower among patients with UEDVT than among those with LEDVT (12% vs. 30%, P = 0.009) (Table 1). In contrast, no significant difference between UEDVT and LEDVT patients was seen regarding carriership of prothrombin G20210A, presence of high FVIII levels, or hyperhomocysteinemia.

We recently found that ex vivo and in vitro thrombin generation (as reflected by plasma D-dimer levels and peak thrombin generation) are independently associated with the risk of recurrent VTE [12,14]. In this study, both D-dimer levels and peak thrombin generation were significantly higher among patients with LEDVT than among those with UEDVT (Table 1).

Recurrent VTE

Recurrent VTE was seen in two of 50 patients with UEDVT (4%) and in 129 of 841 patients with LEDVT (15%). Recurrences in patients with UEDVT were idiopathic and occurred in the ipsilateral arm without concomitant PE. In patients with LEDVT, two of the 129 recurrences were fatal and 12 recurrences (9%) were associated with a temporary risk factor such as surgery or trauma.

As shown in the Kaplan–Meier analysis, the cumulative probability of recurrence was significantly lower among patients with UEDVT than among those with LEDVT (Fig. 1). After 5 years, the likelihood of recurrence was 2% [95% confidence interval (CI) 0–6] among patients with UEDVT as compared with 19% (95% CI  16–22; P = 0.02) among those with LEDVT. As compared with patients with LEDVT, the relative risk of recurrence was 0.23 (95% CI   0.06–0.93; P = 0.039) among UEDVT patients before and 0.26 (95% CI  0.06–1.05; P = 0.059) after adjustment for age, sex, FV Leiden, prothrombin G20210A, hyperhomocysteinemia and high FVIII level.

Figure 1.

 According to Kaplan–Meier analysis, the risk of recurrent venous thromboembolism (VTE) was significantly lower in patients with upper extremity deep vein thrombosis (UEDVT) than in those with lower extremity deep vein thrombosis (LEDVT).


Our study shows that idiopathic UEDVT is a rare manifestation of VTE. Of 3322 patients with all-cause VTE screened for eligibility for AUREC, idiopathic UEDVT was recorded in only 50 (1.5%). This finding compares well with the results from a large US thrombosis registry [1]. Joffe et al. reported that of 5388 DVT patients evaluated, only 268 (5.0%) had UEDVT not associated with central venous catheter insertion. It is of note that in our study population, which consisted of patients with idiopathic rather than secondary DVT, the majority of patients had thrombi located in the deep veins of the leg, whereas only 5.6% of patients had UEDVT.

The large discrepancy between frequencies of idiopathic UEDVT and LEDVT led us to investigate the possibility of different pathomechanisms being accountable for the two conditions. Both differences in gravitational strain and vein valve anatomy are plausible explanations for this phenomenon. We surmised that UEDVT patients might differ from LEDVT patients also in various other respects, including clinical patient characteristics or carriership of congenital clotting abnormalities. Our study is the first to directly compare patients with idiopathic UEDVT and idiopathic LEDVT with respect to clinical and laboratory thrombophilia, and we also determined blood levels of coagulation activation markers in the two patient groups.

Patients with idiopathic UEDVT were more than 10 years younger than patients with LEDVT at manifestation, and they were also significantly slimmer. This observation is in agreement with those of other investigators [1,3], who reported that advancing age and overweight, two of the most common risk factors for LEDVT, do not predispose to UEDVT. UEDVT associated with strenuous activities of the upper arm, which has been shown to approximately double the risk of UEDVT [18], was found in a fifth of our patients. This is in accordance with the findings of Martinelli et al. [4] and Heron et al. [19], who reported similar incidences of exercise-induced UEDVT in their respective cohorts. In our study cohort, UEDVT occurred significantly more frequently in the right arm (62%) than in the left arm, which contrasts with the findings of earlier studies. For instance, van Stralen et al. found that in 62% of non-athletes, UEDVT occurred in the left arm [18], whereas Blom reported a more balanced ratio between left-sided (56%) and right-sided events (44%) [3]. Thoracic outlet syndrome as a possible cause of UEDVT was detected in four patients. Two patients had a cervical rib or a musculoskeletal compression, respectively. None of these patients had a recurrence. As a limitation of our study, a systematic diagnostic work-up was not performed in all patients. Even so, our data suggest that an underlying anatomic compression is present in only a minority of patients with spontaneous UEDVT, and that routine screening by CT scan or MRI is not warranted in this patient population.

FV Leiden and prothrombin G20210A are the most important genetic risk factors for VTE, with incidences in cohorts of thrombosis patients of approximately 30% and 10%, respectively [20]. There is evidence from case–control studies that both FV Leiden and prothrombin G20210A are independent risk factors for idiopathic UEDVT, with odds ratios ranging between 2.5 and 6.2 [3,4,21]. Whereas in our study the frequency of prothrombin G20210A was similar in patients with idiopathic UEDVT and LEDVT, only 12% of UEDVT patients, as compared to 30% of patients with LEDVT, were carriers of FV Leiden. Taking into account that the incidence of FV Leiden in the Austrian population is about 5%, the mutation is probably a risk factor for thrombosis also in our cohort of UEDVT patients. However, the importance of this genetic abnormality appears to be much less in patients with UEDVT than in those with LEDVT. Our finding that the proportion of patients with a family history of VTE was lower among the UEDVT patients than among those with LEDVT also supports the notion that genetic risk factors play a less prominent role in the pathogenesis of UEDVT. We did not find a difference between patients with UEDVT and LEDVT with respect to two other permanent risk factors for DVT, hyperhomocysteinemia and high FVIII level. We cannot comment on antithrombin, protein C and protein S deficiency or the lupus anticoagulant, as patients with such a clotting abnormality were not included in the study.

VTE can be associated with enhanced hemostatic system activation and in vitro thrombin generation [22,23]. We and others have shown that a large proportion of thrombosis patients have increased levels of D-dimer and a peak thrombin level shortly after discontinuation of anticoagulation [12,14,24,25]. These patients have a much higher risk of recurrent VTE, suggesting that ongoing hemostatic system activation and/or an enhanced capacity of the blood to generate thrombin have a role in DVT development. Both D-dimer and peak thrombin were significantly higher in patients with idiopathic LEDVT than in patients with UEDVT. This finding indicates that the hemostatic system is more prominent in LEDVT patients, which may also explain, at least in part, the higher recurrence risk in this patient group.

PE and recurrence are frequent and serious complications of DVT. An important finding of our study was that the proportion of patients with concomitant symptomatic PE was significantly lower among idiopathic UEDVT patients than among patients with LEDVT (8% vs. 31%). Data on the frequency of PE among UEDVT patients are conflicting. In a study from Italy, more than one-third of patients with all-cause UEDVT suffered from concomitant PE [5]. In contrast, none of the 69 patients with UEDVT studied by Spencer et al. had PE [2], and Joffe et al. recorded PE in only 1% of their patient population with UEDVT from all causes [1]. Larger studies on well-defined patient populations are needed in order to better estimate the risk of PE in UEDVT patients. Even so, as compared to LEDVT, UEDVT appears to be a more benign condition in terms of morbidity and mortality.

VTE is a chronic disease with a strong tendency towards recurrence. The likelihood of recurrence among our patients with a first episode of idiopathic LEDVT was as high as 20% after 5 years, and was comparable to what has been found in similar studies [26,27]. Because of the high risk of recurrence, how long patients with idiopathic LEDVT should receive secondary thromboprophylaxis with oral anticoagulants is still a matter of intense debate. With regard to idiopathic UEDVT, only two of our 50 (4%) patients suffered from recurrent DVT. After 5 years, the likelihood of recurrence was 2%, with a 95% CI ranging from 0 to only 6% and was lower than reported by two Italian studies, which showed cumulative probabilities of recurrence of 11% [4] and 7.7% [6] after 5 years, respectively. A ready explanation for this difference is inclusion of patients with a higher risk of recurrence in the two studies. For instance, Martinelli et al. included patients with natural coagulation inhibitor deficiencies or patients with phospholipid antibodies [4], and in the cohort of Prandoni et al., 21% of patients had active cancer [6]. In our cohort, the risk of recurrence in patients with idiopathic UEDVT was almost 80% lower than that in patients with LEDVT. Thus, a longer course of secondary thromboprophylaxis seems not to be justified in this patient group. In addition, because of the low recurrence rate and the lower incidence of genetic risk factors, routine laboratory thrombophilia screening does not appear to be useful.

In conclusion, the pathogenesis and natural course of the disease appear to be different in patients with idiopathic UEDVT and LEDVT. Patients with UEDVT differ from those with LEDVT with regard to the incidence of clinical and laboratory risk factors, including age and the carriage of FV Leiden. UEDVT patients have a lower degree of in vivo hemostatic system activation and in vitro thrombin generation capacity, and are at a much lower risk of concomitant PE and recurrent DVT.

Disclosure of Conflict of Interests

The authors state that they have no conflict of interest.