Risk Factors for Deep Venous Thrombosis of the Upper Extremities

Methods

Patients

Deep Venous Thrombosis of the Upper Extremities

Forty-nine consecutive unrelated patients who had a first episode of deep venous thrombosis of the upper extremities were referred to the Thrombosis Center in Milan, Italy, between January 1993 and January 1996 to be screened for hypercoagulable states. Thirty-six of the patients (15 men and 21 women; median age, 28 years [range, 18 to 56 years]) had primary deep venous thrombosis and were evaluated at least 2 months after the acute thrombotic episode. None had evidence of overt neoplastic, autoimmune, liver, or renal disease; congestive heart failure; or history of drug abuse or clavicular fracture. Anticoagulant therapy had been discontinued at least 1 month before blood sampling in 34 patients, but 2 patients were still receiving anticoagulant agents. Diagnosis had been confirmed by duplex scanning in 19 patients (sensitivity, 100%; specificity, 93% [2]) and by contrast venography in 17 patients. Thrombosis involved the axillary vein, subclavian vein, or both in 34 patients and the brachial vein in 2 patients; in 61% of patients, the thrombus was located in the right arm. No patient had had symptoms of pulmonary embolism.

Deep Venous Thrombosis of the Lower Extremities

From January 1993 to January 1996, 162 patients with a first episode of deep venous thrombosis of the lower extremities were referred to our center to be screened for hypercoagulable states. Of these patients, 121 (48 men and 73 women; median age, 31 years [range, 15 to 61 years]) were retrospectively included in our study because they matched (by age [± 5 years], sex, geographic origin [region of birth], and social status [level of education]) the patients with venous thrombosis of the upper extremities and had no evidence of overt neoplastic, autoimmune, liver, or renal disease. Twenty of the patients were receiving oral anticoagulant agents. Thrombosis was diagnosed by ultrasonography or contrast venography.

Any of the circumstantial factors known to trigger thrombosis of the lower extremities were recorded for both groups of patients. In addition, the association between thrombotic episodes and strenuous muscular activity of the affected extremities was evaluated in all patients; exercise is believed to be a pathogenic factor in patients with deep venous thrombosis of the upper extremities [2, 8-10].

Controls

One hundred eight healthy persons (45 men and 63 women; median age, 31 years [range, 19 to 58 years]) were enrolled as controls because they matched patients with venous thrombosis of the upper extremities for the following variables and risk factors: age (± 5 years), sex, geographic origin, and social status. The controls were friends or partners of the patients and were not biologically related to the patients. Previous occurrence of thrombosis was excluded by a structured questionnaire that has been validated for retrospective diagnosis of thrombosis [11]. None of the controls had been exposed to any of the triggering factors for 2 weeks before the visit; female controls using contraceptives continued to do so.

Laboratory Tests

Resistance to activated protein C was assayed by a clotting method based on the activated partial thromboplastin time [12]. When test results were abnormal or borderline, DNA analysis for 1691G→A substitution in coagulation factor V gene was done [13]. Antithrombin, protein C, and protein S levels were measured as described elsewhere [14]. Presence of antiphospholipid antibodies was confirmed if lupus anticoagulant, anticardiolipin antibodies, or both were present. Total plasma homocysteine levels were measured by high-performance liquid chromatography and electrochemical detection [15] after patients had fasted overnight and 4 hours after they had received an oral methionine load (3.8 g/m² body surface area). Hyperhomocysteinemia was diagnosed if fasting levels or absolute increments of the postmethionine load above fasting levels exceeded the 95th percentile of distribution values in healthy controls. Because nutritional deficiencies can cause high homocysteine levels [16], serum vitamin B₁₂ and folate levels were also measured.

Plasma homocysteine levels were not measured in 4 patients who had deep venous thrombosis of the lower extremities. The 22 patients receiving oral anticoagulant treatment (2 patients with upper-extremity and 20 patients with lower-extremity venous thrombosis) were included in the study. These patients were tested only for antithrombin, plasma homocysteine, antiphospholipid antibodies, and factor V mutation because measurements of vitamin K-dependent protein C and protein S levels and resistance to activated protein C were unreliable.

Statistical Analysis

We used the Student t-test to compare the age of patients at the time of thrombosis. Because use of oral contraceptives and pregnancy exclude each other, the frequency of women receiving oral contraceptives was calculated after exclusion of pregnant women and the frequency of pregnancy was calculated after exclusion of women receiving oral contraceptives. Odds ratios and 95% CIs were calculated to compare the risk for deep venous thrombosis of the upper extremities with that of the lower extremities. Adjustment for matching factors and potential confounding factors was done by conditional logistic regression analysis or the Mantel-Haenszel procedure.

Results

The overall prevalence of abnormalities of the natural anticoagulant system (deficiencies of antithrombin, protein C, or protein S and resistance to activated protein C) in patients who had deep venous thrombosis of the upper extremities (9%) was similar to that in controls (6%) (odds ratio, 2.6 [95% CI, 0.4 to 17.1]) but was significantly lower than that in patients who had venous thrombosis of the lower extremities (31%) (odds ratio, 0.1 [CI, 0.06 to 0.6]). In contrast, the prevalence of abnormalities was higher in patients who had deep venous thrombosis of the lower extremities than in controls (odds ratio, 13.3 [CI, 3.0 to 58.9]) (Table 1).

The prevalence of hyperhomocysteinemia was similar in patients with deep venous thrombosis of the upper extremities (6%) and controls (7%) (odds ratio, 0.8 [CI, 0.1 to 7.9]) but was higher in patients with thrombosis of the lower extremities than in controls (14%) (odds ratio, 2.5 [CI, 0.8 to 7.8]); however, the difference was not statistically significant (Table 1). Three patients with thrombosis of the lower extremities and two controls had hyperhomocysteinemia and low serum vitamin B₁₂ levels, folate concentrations, or both (data not shown). Antiphospholipid antibodies were found in eight patients (7%) who only had deep venous thrombosis of the lower extremities.

The overall prevalence of abnormalities was calculated for those patients and controls who had a complete laboratory screening. Prevalences were similar in patients who had deep venous thrombosis of the upper extremities (15%) and controls (12%) (odds ratio, 1.6 [CI, 0.4 to 7.3]). However, the overall prevalence in patients who had thrombosis of the lower extremities was much higher than that in patients who had thrombosis of the upper extremities (42%) (odds ratio, 0.1 [CI, 0.1 to 0.5]) and much higher than that in healthy controls (odds ratio, 14.7 [CI, 5.3 to 40.8]) (Table 1).

The overall prevalence of triggering factors was significantly lower in patients who had thrombosis of the upper extremities (39%) than in those who had thrombosis of the lower extremities (69%) (odds ratio, 0.1 [CI, 0.04 to 0.3]) (Table 2). The difference was limited to women and was attributable to use of oral contraceptives (19% of patients with thrombosis in an upper extremity and 53% of patients with thrombosis in a lower extremity) (odds ratio, 0.3 [CI, 0.1 to 0.9]) and to pregnancy or puerperium (0% and 33%, respectively). Thirty percent of female controls were using oral contraceptives; none was pregnant or postpartum. The frequency of triggering factors in women receiving oral contraceptives was not significantly different between women who had thrombosis of the upper extremities and controls (19% and 30%, respectively). These percentages were significantly lower than those in women who had thrombosis of the lower extremities (53%) (odds ratio, 0.2 [CI, 0.1 to 0.7] and 2.1 [CI, 1.0 to 4.4], respectively). In men, the overall prevalence of triggering factors was similar among patients who had deep venous thrombosis of any extremity (47% and 52%, respectively; odds ratio, 0.9 [CI, 0.3 to 2.5]) (Table 2). Men who had thrombosis of the lower extremities had been more frequently exposed to surgery and trauma or immobilization than those who had thrombosis of the upper extremities (17% and 27% compared with 0% and 13%, respectively). The most prevalent triggering factor in men who had thrombosis of the upper extremities was strenuous muscular activity, such as boxing or weight lifting. In contrast, the factor most prevalent in men with lower-extremity thrombosis was less strenuous muscular activity, such as mountain biking or soccer (33% compared with 8%; odds ratio, 4.8 [CI, 0.8 to 30.3]).

Discussion

In this study, we found that a hypercoagulable state caused by abnormalities of the anticoagulant system or of homocysteine metabolism did not occur often in patients who had primary deep venous thrombosis of an upper extremity. In comparison, the prevalence of hypercoagulable states was relatively high in a similar group of patients who had deep venous thrombosis of the lower extremities, thus confirming observations documented elsewhere [17-20]. Another important difference between the two sets of patients was the frequency of the association between thrombotic episodes and conditions that temporarily cause hypercoagulability, such as oral contraceptive intake and pregnancy or puerperium; the frequency was high only in women whose lower extremity was affected. In men whose upper extremity was affected, the most frequent triggering factor was strenuous muscular activity of the upper extremity, a finding that has been documented elsewhere [2, 8-10]; the most common factors in men with an affected lower extremity were surgery, trauma, or immobilization.

Limitations of our study might be the small number of patients who had primary deep venous thrombosis (which can be attributed to the relative rarity of the condition) and the possibility of recall bias when patients reported triggering factors. However, we recorded triggering factors within 2 weeks before the study visit. In addition, the potential for patients to underreport such factors as surgery, trauma, or pregnancy and puerperium is limited. Perhaps the potential for patients to underreport may exist for a factor such as strenuous muscular activity of the limbs. Despite these limitations, the study clearly indicates that the pathogenesis of deep venous thrombosis of the upper extremities differs from that of deep venous thrombosis of the lower extremities and that mechanisms other than a procoagulant imbalance prevail. Hemodynamic factors and vascular damage probably play a major role. Because the upper extremities lack valve pockets, which represent an area of maximum stasis, local accumulation of activated coagulation factors and fibrin deposition are less likely to occur. However, in persons with such predisposing anatomic abnormalities as thoracic outlet compression or anomalous musculofascial bands, which are often unappreciated and difficult to diagnose, strenuous muscular activity may aggravate the intrinsic compression of the veins. Such aggravation could cause endothelial damage, fibrosis around the vein, and reduction of blood flow, any of which could contribute to formation of thrombi.

In conclusion, our study indicates that, unlike patients who have deep venous thrombosis of the lower extremities, patients with primary thrombosis of the upper extremities should not be investigated for congenital or acquired abnormalities that cause hypercoagulability. Rather, clinicians should look for a recent history of strenuous muscular activity of the affected extremity. Further studies on a large number of patients are necessary to confirm our data and to clarify whether surgical correction or anticoagulant treatment [8-10] can effectively prevent recurrences.

Dr. Taioli: Epidemiology Unit, Istituto di Ricovero e Cura a Carattere Scientifico Maggiore Hospital, University of Milan, via F. Sforza 35, 20122 Milan, Italy.