Central venous catheters and upper-extremity deep-vein thrombosis complicating immune heparin-induced thrombocytopenia

Heparin-induced thrombocytopenia (HIT) is an acute, transient prothrombotic disorder caused by heparin-dependent, platelet-activating antibodies.1-3 HIT can be regarded as a hypercoagulability state, based upon the greatly elevated levels of thrombin-antithrombin complexes4,5 (a marker of in vivo thrombin generation) observed in patients with HIT, as well as the strong association between HIT and venous and arterial thrombosis.1-3 Indeed, the most common thrombotic manifestation of HIT is lower-extremity deep-vein thrombosis (DVT).2,3,6,7 However, previous studies have not examined whether HIT is also associated with upper-extremity DVT. This is an important question, because it is known that upper-extremity DVT can complicate chronic hypercoagulability states such as antithrombin deficiency and the antiphospholipid antibody syndrome.8-10Further, upper-extremity DVT also is known to be associated with central venous catheter (CVC) use.8-10 Since HIT typically occurs in hospitalized patients receiving heparin who may also have been treated with a CVC, addressing this question would provide the opportunity to study the interaction of a systemic hypercoagulability state (HIT) with a localizing factor (CVC use) in explaining HIT-associated upper-extremity DVT.

We reviewed 260 consecutive cases of serologically confirmed HIT diagnosed in one of 4 academic hospitals in Hamilton, ON, Canada, over an 18-year period ending June 30, 1998. All patients had a positive platelet ¹⁴C-serotonin release assay: 243 patients had a 50% or greater fall in the platelet count,3 and 17 patients had a smaller fall in the platelet count, accompanied by thrombotic or other typical sequelae of HIT, for example, skin lesions at heparin-injection sites.11 12 Of these 260 patients, 145 had a CVC placed during the 2-week period prior to the episode of HIT. The remaining 115 patients with HIT did not receive a CVC.

Two non-HIT control groups were identified from studies performed in the same medical community that met all of the following criteria: a high rate of CVC use; systematic surveillance for lower-limb DVT using either contrast venography or compression ultrasonography; systematic platelet count monitoring; and serological investigations for HIT antibodies (using the platelet ¹⁴C-serotonin release assay) in patients with clinically suspected HIT. Control group 1 consisted of patients without HIT identified in a study of postoperative antithrombotic heparin prophylaxis following major orthopedic surgery (hip arthroplasty), as described.2Control group 2 consisted of patients without HIT identified in a study13 of intensive-care unit patients in whom systematic screening for DVT (twice-weekly compression ultrasonography) was performed. Imaging investigations for upper-extremity DVT were performed in HIT patients and controls only if signs or symptoms of upper-extremity venous thrombosis were present.

All DVTs in HIT patients and in non-HIT controls were confirmed using either contrast venography or compression ultrasonography. We included all lower-extremity DVTs involving popliteal or more proximal veins and all upper-extremity DVTs involving at least one of the brachial, axillary, subclavian, or internal jugular veins. The reason for including only proximal lower-limb DVTs was because only control group 1 had undergone screening with contrast venography, thus permitting routine identification of calf DVT only in this group. Additionally, HIT has been shown to be strongly associated with proximal lower-limb DVT, but not with DVT that involves only the calf veins.2 

For the patients with HIT, we considered that a DVT was associated with their episode of HIT if it occurred 5 or more days after starting heparin and was associated with a platelet count fall that was linked serologically to HIT antibodies.2 We further recorded whether the DVTs involved the right or left extremities, as well as the timing of insertion and removal of the CVC in relation to the episode of upper-limb DVT. The test of agreement for right versus left location of CVC use and location of upper-extremity DVT was performed using the kappa statistic. Comparisons between groups were made using χ² Fisher exact test (2-sided). The odds ratio was calculated using the Mantel-Haenszel statistic.

Table 1 shows that patients with HIT were significantly more likely to develop either upper-extremity DVT or proximal lower-extremity DVT compared with either control group of patients without HIT. Whereas 14 of 260 (5.4%) patients with HIT developed upper-extremity DVT, this complication was not seen in any of the 637 postoperative orthopedic surgery patients (P < .0001) and in only 3 (1.1%) of the 261 intensive-care unit patients (P = .0066). Whereas 42.3% of HIT patients developed proximal lower-limb DVT, only 4.1% of postoperative orthopedic surgery patients and 9.6% of intensive-care unit patients developed proximal lower-limb DVT (P < .0001 for both comparisons).

CVC use was a crucial factor in explaining an increased risk of upper-limb DVT, which occurred in 14 of 145 (9.7%, 95% CI, 5.4-15.7) patients with both HIT and CVC use, compared with none of the 115 (0%, 95% CI, 0.0-3.2) HIT patients without a CVC (P = .000 35). In contrast, use of a CVC did not confer an increased risk of lower-extremity proximal DVT in either the HIT or the control populations without HIT. Among all HIT and control patients who had CVC use, the frequency of upper-extremity DVT was significantly higher in the patients with HIT: 14 of 145 (9.7%) compared with 3 of 484 (0.6%, 95% CI, 0.1-1.8); odds ratio, 17.1 (4.9-60.5;P < .0001).

Table 2 shows the influence of the side of placement of the CVC (right versus left upper-extremity placement) in determining the location of the upper-extremity DVT. All 14 upper-extremity DVTs occurred at the CVC site (right, 12; left, 2; kappa = 1.0; P = .011). In 7 patients, the DVT became clinically evident when the CVC was still indwelling. For the remaining 7 patients, the DVT became clinically evident a median of 4 days (range, 2-10 days) following removal of the CVC, indicating that the localizing prothrombotic effect of the intravascular catheter persists for a time, most likely by residual injury to the upper-extremity vessel resulting from the recent placement of the CVC. This represents another way that CVC use can influence clinical outcome in HIT besides its known role as a potential source for heparin exposure.11 

Although we used a retrospective case-control study design, we believe our study was unlikely to have suffered from ascertainment bias, for example, because physicians may have been more likely to diagnose upper-extremity DVT in a patient with acute HIT. In our patients, all upper-limb imaging studies were performed because of signs or symptoms of upper-limb DVT, and none were obtained because of routine screening of upper-limb vessels for DVT in asymptomatic patients with HIT. Further, our control patients were identified from 2 prospective clinical studies in which all patients underwent daily clinical assessment for thrombosis.

In conclusion, our study shows that HIT is strongly associated with both upper-extremity and lower-extremity DVT (Table 1), consistent with the view of HIT as a hypercoagulability state.4 5 However, our study also has identified a strong additional factor, namely the localizing influence of a CVC, in determining the occurrence of upper-extremity DVT among patients with HIT, and particularly in determining the site of the upper-limb DVT. Indeed, in all 14 patients with HIT-associated upper-extremity DVT, the DVT occurred in the same limb in which the CVC had been placed. These data indicate that while HIT confers a systemic prothrombotic risk—as manifested by increased risk for both lower- and upper-extremity DVT—the localization of such thrombosis to upper-extremity veins is strongly influenced by CVC use.