The Pathogenesis of Venous Limb Gangrene Associated with Heparin-Induced Thrombocytopenia

Methods

Case-Control Studies

In the first of two case–control studies, we compared patients with heparin-induced thrombocytopenia who developed venous limb gangrene with patients with heparin-induced thrombocytopenia who developed acute limb arterial thrombosis. In the second case–control study, we compared patients who developed venous limb gangrene with patients who did not develop venous limb gangrene during warfarin treatment of heparin-induced thrombocytopenia and deep venous thrombosis.

Patients

Case-patients and controls were identified from review of the clinical and laboratory records of all 158 patients treated for heparin-induced thrombocytopenia (platelet count nadir, <150 × 10⁹/L) in one of the five Hamilton, Ontario, hospitals during a 15-year period that ended on 31 December 1994. The diagnosis of heparin-induced thrombocytopenia was confirmed serologically in all patients [17, 18]; 133 patients had been included in previous studies [3, 19].

Four patient summaries are included in this report; three of the four patients had venous limb gangrene (patients 1, 2, and 3), and one patient had severe venous limb ischemia that resolved on reversal of warfarin treatment with vitamin K and plasmapheresis (patient 4).

Definitions

Limb arterial thrombosis was defined as abrupt arterial occlusion of a limb with absent pulses and was verified in all patients by angiography or surgical thrombectomy. Venous limb gangrene was defined as distal ischemic tissue necrosis complicating deep venous thrombosis despite palpable or Doppler-identifiable arterial pulses. Deep venous thrombosis was confirmed by contrast venography or duplex compression ultrasonography in all patients. All pathologic material was reviewed to evaluate the types of vessels involved in the thrombotic process. Central skin necrosis was defined as skin necrosis that occurred in the breast, abdomen, buttocks, or thigh in association with warfarin treatment.

Coagulation Studies

Serial citrated plasma samples were available from 34 consecutive patients with heparin-induced thrombocytopenia who were treated at one hospital over a 2-year period. These samples were collected and tested according to a local study protocol approved by the hospital's institutional review board. Before testing, all samples were stored in aliquots frozen at −70°C. Four of the 34 patients (including patients 2, 3, and 4) developed venous limb gangrene or severe venous limb ischemia; this allowed us to test the hypothesis that warfarin therapy could contribute to venous limb gangrene. Control studies were done by using plasma obtained during a clinical trial of heparin prophylaxis [3] from 59 patients with deep venous thrombosis but not heparin-induced thrombocytopenia.

We used one-stage assay techniques to measure levels of the following vitamin K-dependent procoagulant factors: prothrombin (factor II), factor VII, and factor X [20]. Protein C activity was determined by using a functional assay from Diagnostica Stago (Wellmark Diagnostics Ltd., Guelph, Ontario, Canada). Free protein S levels were measured by precipitation of the protein S bound to C4b-binding protein with polyethylene glycol [21]; the supernatant that contained the free protein S was measured by enzyme-linked immunosorbent assay [22] using antibodies to protein S (Affinity Biologicals, Hamilton, Ontario, Canada). Thrombin-antithrombin complex levels were determined by using an enzyme-linked immunosorbent assay [23] (Behring Diagnostics, Montreal, Quebec, Canada) to evaluate in vivo thrombin generation [24]. Antithrombin was measured by use of a chromogenic factor Xa inhibition method (Chromogenix, Helena Laboratories, Mississauga, Ontario, Canada) [25]. Evaluation for factor V Leiden was performed by using a functional assay for resistance to activated protein C [26] and by direct demonstration of the mutation at the DNA level [27].

Statistical Analysis

We used both quantitative and qualitative measures to compare groups. Because quantitative measures tended to have skewed distributions, medians rather than means were used as summary statistics and in comparisons between groups. The Mann-Whitney test was used to compare medians, and associated nonparametric methods [28] were applied to estimate 95% CIs on differences between medians (MINITAB Release 10.5, Xtra software, Minitab, Inc., State College, Pennsylvania). Binary variables were summarized as proportions and were compared between groups by using the Fisher exact test [29]. Differences between groups for the binary variables were represented as risk ratios with CIs according to the method of Thomas [30]. For cases in which the risk ratio was zero or infinite, StatXact (Cytel Software Corp., Cambridge, Massachusetts) was used to calculate the exact 95% CIs for the odds ratios. The nonzero upper or lower bound that StatXact computed was then used to produce a quadratic Equation that, when solved, provided the expected event count for the first comparison group. The corresponding upper or lower bound for the risk ratio was then calculated by recomputing the 2 × 2 frequency table.

Selected Case Reports

Patient 1

A 56-year-old woman developed heparin-induced thrombocytopenia (platelet count nadir, 37 × 10⁹/L) that was recognized on day 8 of intravenous heparin therapy given for bilateral, idiopathic proximal deep venous thrombosis of the lower limbs and pulmonary embolism. Heparin therapy was discontinued that day, and 20 mg of warfarin was given daily for 2 consecutive days. The following day, the patient had an international normalized ratio (INR) of 9.4 and developed necrosis involving eight toes, despite palpable pedal pulses (Figure 1). The patient was managed conservatively; sloughing of the gangrenous tissues was followed by healing without the need for amputation.

Patient 2

A 49-year-old woman received treatment with heparin and warfarin for 19 days because of idiopathic proximal deep venous thrombosis of the left lower limb. At that time, progressive symptoms of pain and swelling in the left lower limb prompted measurement of the platelet count, which was found to be 69 × 10⁹/L (nadir); heparin therapy was discontinued. The INR was therapeutic at 2.2, and warfarin, 7.5 mg/d, was given for the next 2 days. The platelet count returned to normal within 4 days of discontinuation of heparin therapy. However, venous limb gangrene involving the distal left foot was first seen on the third day after heparin therapy ended (Figure 2 A). The INR was 7.2, protein C activity was less than 0.01 U/mL (normal, 0.65 to 1.29 U/mL), and free protein S level was markedly reduced (0.06 U/mL; normal, 0.24 to 0.62 U/mL). The prothrombin level was only moderately reduced (0.26 U/mL; normal, 0.5 to 1.6 U/mL), and thrombin-antithrombin complex levels were elevated (16.6 ng/mL; normal, <4.2 ng/mL). The patient required a below-the-knee amputation. Occlusive venous thrombi were noted in small venules and medium-sized veins, and arteries were normal (Figure 2 B, C, and D).

Patient 3

A 35-year-old woman received heparin prophylaxis for injuries sustained in a motor vehicle accident. She had no fractures or soft tissue injury to her left lower limb; she developed proximal deep venous thrombosis of the left lower limb on day 8 of subcutaneous heparin prophylaxis in association with a decrease in platelet count from 216 × 10 (⁹)/L (day 6) to 144 × 10⁹/L (day 7). however, heparin-induced thrombocytopenia was not suspected, and intravenous therapeutic-dose heparin was started; the platelet count measured 3 days later was 30 × 10 (⁹)/L (platelet count nadir, 20 × 10⁹/L).

Heparin therapy was discontinued, and a defibrinogenating agent (ancrod) was given. In addition, 15 mg of warfarin was given each day for 3 days. The INR increased to 7.2, and venous gangrene involving the entire left foot was seen. The gangrene necessitated below-the-knee amputation. Occlusive thrombi were seen in the small venules and medium-sized veins. Four days after warfarin therapy was started, the patient developed a 5 × 5 cm area of skin necrosis in the left lower quadrant of the abdomen. Warfarin therapy was discontinued, and ancrod treatment was continued until thrombocytopenia resolved. Warfarin therapy was then resumed, without adverse consequences. The coagulation results in this patient are summarized later in this article.

Patient 4

A 34-year-old woman developed heparin-induced thrombocytopenia (platelet count nadir, 20 × 10⁹/L) complicated by deep venous thrombosis of the right lower limb on day 7 of subcutaneous heparin prophylaxis after injuries sustained in a motor vehicle accident. Heparin therapy was discontinued, and the patient received ancrod and warfarin. Two days later, when the platelet count was 25 × 10⁹/L, thrombosis of the left axillary vein developed. Four days later, when the platelet count was 68 × 10⁹/L and the INR had increased to 4.0, pain, swelling, and cyanosis affecting the left hand developed despite palpable radial and ulnar pulses. Warfarin therapy was discontinued, vitamin K was given, and the patient underwent plasmapheresis using fresh frozen plasma replacement. Symptoms and signs of ischemia of the hand rapidly resolved. Plasma obtained immediately before apheresis showed the following measures: protein C activity, 0.04 U/mL (normal, 0.65 to 1.29 U/mL); free protein S level, 0.07 U/mL (normal, 0.24 to 0.62 U/mL); prothrombin level, 0.31 U/mL (normal, 0.5 to 1.6 U/mL); factor VII level, 0.12 U/mL (normal, 0.5 to 1.6 U/mL); factor X level, 0.38 U/mL (normal, 0.5 to 1.6 U/mL); and thrombin-antithrombin complex level, 17.1 ng/mL (normal, <4.2 ng/mL). After apheresis, these abnormalities were corrected.

Clinical and Laboratory Studies

Patients

We identified 158 patients with serologically confirmed heparin-induced thrombocytopenia. The indications for heparin therapy in these patients were treatment of deep venous thrombosis or pulmonary embolism (n = 22), treatment of acute myocardial infarction or unstable angina (n = 14), antithrombotic prophylaxis in trauma patients or patients having surgery (n = 91), antithrombotic prophylaxis in medical patients (n = 21), and miscellaneous indications (such as use in invasive catheters) (n = 10).

From the 158 patient records, we identified 18 patients with ischemic limb necrosis associated with heparin-induced thrombocytopenia. Ten of these patients had acute arterial thrombosis that involved 12 lower limbs. Limb amputation was necessary in 3 of these patients, all of whom died; surgical thrombectomy resulted in limb salvage in the remaining 7 patients, all of whom survived. Eight patients had venous limb gangrene involving 10 lower limbs. All 8 of these patients had acute symptomatic deep venous thrombosis that was radiologically confirmed before venous gangrene developed.

In 5 patients, deep venous thrombosis was the initial reason for heparin use; in the other 3 patients, deep venous thrombosis occurred during heparin-induced thrombocytopenia. In each instance, venous limb gangrene occurred in the limb that was affected by concurrent symptomatic deep venous thrombosis. Two patients with extensive bilateral lower-limb venous gangrene died, 2 required below-the-knee amputation, 2 required amputations of distal portions of one or both feet, and 2 had distal skin sloughing but did not require surgical amputation.

Venous ischemia of the hand resolved in one patient (patient 4) after reversal of warfarin anticoagulation, and no tissue was lost. This patient is not included in the group of 8 patients with venous limb gangrene.

Clinical Profile of Venous Limb Gangrene

Comparison with Patients Who Had Limb Arterial Thrombosis

Table 1 and Table 2 contrast the clinical and laboratory profiles of the 10 patients who developed limb arterial thrombosis with those of the 8 patients who developed venous limb gangrene.

Arterial ischemia usually occurred as the initial manifestation of heparin-induced thrombocytopenia. In contrast, all 8 patients with venous limb gangrene had heparin-induced thrombocytopenia and deep venous thrombosis followed by development of venous limb gangrene after heparin therapy was discontinued and treatment with warfarin was either initiated or continued.

Comparison with Patients Who Received Warfarin but Did Not Develop Gangrene

The 158 patients with heparin-induced thrombocytopenia included 66 patients with deep venous thrombosis who received warfarin; 8 (12%) of these developed venous limb gangrene. The median peak INR was 5.8 in the 8 patients who developed venous limb gangrene and 3.1 in the 58 patients who did not develop venous limb gangrene, a difference of 2.7 (95% CI, 1.5 to 4.3; P < 0.001). There was a trend toward a higher median dose of warfarin being given over the first 5 days to patients who developed venous limb gangrene (Table 1).

Coagulation Studies

Included in the 158 patients were 34 consecutive patients who were treated for heparin-induced thrombocytopenia at one hospital and who had plasma samples available for further study. The median thrombin-antithrombin complex levels were 43.2 ng/mL (range, 2.1 to 225 ng/mL) in the 34 patients with heparin-induced thrombocytopenia and 7.6 ng/mL (range, 1.9 to 26.0 ng/mL) in the 59 patients with deep venous thrombosis without heparin-induced thrombocytopenia (difference, 35.4 ng/mL [CI, 18.1 to 50.3 ng/mL]; P < 0.001). Protein C activity was within the normal range (0.65 to 1.29 U/mL) in most patients with heparin-induced thrombocytopenia when they were not receiving warfarin therapy (Figure 3). Free protein S levels were also within the normal range (0.24 to 0.62 U/mL) in most of these patients (88% [22 of 25]). Only one patient with heparin-induced thrombocytopenia had overt disseminated intravascular coagulation at baseline, as defined by hypofibrinogenemia (plasma fibrinogen level < 1.5 g/L); this patient did not receive warfarin and did not develop venous limb gangrene.

Twenty of the 34 consecutive patients (59%) with heparin-induced thrombocytopenia received warfarin during treatment of heparin-induced thrombocytopenia. Three of these 20 patients (including patients 2 and 3) developed venous limb gangrene. Another patient developed venous ischemia of the hand that responded to reversal of warfarin anticoagulation (patient 4). Most patients with heparin-induced thrombocytopenia who were treated with warfarin had elevated thrombin-antithrombin complex levels and decreased protein C activity during warfarin treatment. In the four patients who developed either venous limb gangrene or severe ischemia, however, the ratio of thrombin-antithrombin complex to protein C activity during warfarin treatment was higher than in patients receiving warfarin who did not develop venous ischemic complications (Figure 3).

Figure 4 shows the serial platelet counts, thrombin-antithrombin complex levels, prothrombin levels, and functional protein C activity in patient 3. Venous limb gangrene and central skin necrosis occurred during the first course of warfarin therapy; at this time, thrombocytopenia persisted, thrombin generation remained elevated, and protein C activity decreased. After heparin-induced thrombocytopenia resolved, warfarin was readministered without complications.

Evaluation for Congenital Hypercoagulable States

Blood samples were available at baseline or follow-up for 5 of the 8 patients who developed venous limb gangrene (including patients 2 and 3) and in the patient who developed venous limb ischemia that resolved on reversal of warfarin anticoagulation (patient 4). Protein C activity, free protein S levels, and antithrombin levels were normal in these six patients; no patient had resistance to activated protein C (factor V Leiden mutation).

Discussion

Acute arterial occlusion by platelet-rich thrombi (“white clot syndrome”) is the classic complication of heparin-induced thrombocytopenia [1, 5-11]. In this report, we described eight patients with heparin-induced thrombocytopenia who had a different type of ischemic limb event. All eight patients developed distal ischemic necrosis of a limb that accompanied deep venous thrombosis during treatment of heparin-induced thrombocytopenia (venous limb gangrene). In each patient, venous limb gangrene developed after heparin therapy had been discontinued and during anticoagulation with warfarin.

Several clinical and laboratory observations support a pathogenic role for warfarin in the progression to venous limb gangrene in these patients. First, patients with heparin-induced thrombocytopenia who developed venous limb gangrene were more likely to have been receiving warfarin at the onset of limb ischemia than were patients with heparin-induced thrombocytopenia who developed limb arterial thrombosis. Second, the apparent anticoagulant effect of warfarin was greater in patients with venous limb gangrene, as measured by the median INR, than in patients who did not develop venous limb gangrene during warfarin treatment of deep venous thrombosis.

Third, plasma samples from patients with venous limb gangrene had evidence of greater in vivo thrombin generation (higher levels of thrombin-antithrombin complexes) for a corresponding level of protein C activity than did samples from controls (Figure 3 and Figure 4). This hemostatic profile suggests that the remaining protein C activity may be insufficient to regulate the increased thrombin generation occurring in these patients. Finally, the resolution of severe venous limb ischemia in patient 4 after reversal of warfarin anticoagulation with vitamin K and plasma is also consistent with our hypothesis.

In certain clinical circumstances, particularly congenital heterozygous deficiency of protein C [13-16], administration of warfarin can lead to an unusual thrombotic syndrome known as warfarin-induced skin necrosis [31-34]. These patients usually develop central skin necrosis involving the breast, buttocks, abdominal wall, or anterior surface of the thighs; necrosis of the feet or hands occurs in about 10% of patients [33, 34]. Congenital deficiency of protein S has also been implicated in warfarin-induced skin necrosis [35, 36]. Theoretically, resistance to activated protein C related to inheritance of factor V Leiden (which reduces the ability of activated protein C to degrade factor Va [27, 37]) might also predispose patients to thrombotic sequelae of warfarin administration. However, none of our patients with venous limb gangrene was found to have a hereditable abnormality in the protein C anticoagulant pathway. Furthermore, only one of the eight patients with venous limb gangrene in our study developed warfarin-induced central skin necrosis. Thus, the syndrome of venous limb gangrene observed in our patients differs somewhat from classic warfarin-induced skin necrosis.

Recent studies suggest a possible mechanism by which patients with heparin-induced thrombocytopenia might be predisposed to develop warfarin-induced venous limb gangrene. Heparin-induced thrombocytopenia is associated with proximal deep venous thrombosis [3], either because heparin treatment for initial deep venous thrombosis is complicated by heparin-induced thrombocytopenia or because deep venous thrombosis occurs as a thrombotic complication of heparin-induced thrombocytopenia. Thus, patients with heparin-induced thrombocytopenia often have clinically active deep venous thrombosis when they are receiving warfarin.

We believe that this syndrome could be related to the unique prothrombotic state associated with heparin-induced thrombocytopenia. We have reported that IgG from patients with heparin-induced thrombocytopenia stimulates the formation of platelet-derived microparticles with procoagulant activity [38]. Procoagulant microparticles have been shown to stimulate both procoagulant (prothrombinase) and anticoagulant (activated protein C-mediated degradation of factor Va) pathways [38-40]. We have also observed increased circulating levels of platelet-derived microparticles in patients with heparin-induced thrombocytopenia [38]. These microparticles could explain the marked in vivo thrombin generation seen in our patients, which in turn could predispose patients to an imbalance of procoagulant and anticoagulant processes: increased thrombin generation leading to increased generation and consumption of activated protein C, combined with warfarin-induced deficiency of a key thrombin regulatory zymogen, protein C.

We believe that several reasons explain why the possible etiologic role of warfarin in venous limb gangrene in patients with heparin-induced thrombocytopenia is not widely recognized. First, it is known that heparin-induced thrombocytopenia itself is related to limb loss, usually in association with limb arterial thrombosis [1, 6-11]; when this knowledge is considered with our observation that a clinically important thrombosis (deep venous thrombosis) was invariably present before the use of warfarin, clinicians might readily ascribe the subsequent venous limb gangrene to the natural history of heparin-induced thrombocytopenia rather than to a consequence of its treatment. Second, these patients develop an atypical clinical profile for warfarin-induced skin necrosis (predominant distal ischemic necrosis of a limb rather than central skin necrosis; low frequency of hereditable abnormalities of the protein C anticoagulant pathway). These atypical clinical and laboratory features would make the association between venous limb gangrene and warfarin use less readily apparent. Third, the proportion of patients receiving warfarin for heparin-induced thrombocytopenia and deep venous thrombosis who develop venous limb gangrene is relatively small (approximately 12% in our series). Ironically, these patients are usually considered “well anticoagulated” because their INRs were invariably higher than the usual therapeutic range for warfarin (INR, 2.0 to 3.0). However, the comparison of plasma from four of our affected patients with that obtained from the control patients revealed that the high INRs corresponded paradoxically to a “procoagulant” state because thrombin generation persisted despite warfarin use but was complicated by a critical reduction in protein C activity (that is, a disturbance in procoagulant-anticoagulant balance).

Most of our patients with arterial occlusion associated with heparin-induced thrombocytopenia underwent limb-saving surgical thrombectomy. Unfortunately, no surgical procedure is available for small-vessel venous thrombi. This may explain why venous limb gangrene was responsible for more than half of the limb loss associated with heparin-induced thrombocytopenia in our series. Our data suggest that the use of warfarin to treat deep venous thrombosis in some patients with heparin-induced thrombocytopenia may lead to venous limb gangrene as a result of the interaction of the effects of two anticoagulant agents: an initial idiosyncratic reaction (heparin-induced thrombocytopenia) that, in turn, increases the thrombotic risk for a nonidiosyncratic disturbance in procoagulant-anticoagulant hemostatic balance (warfarin-induced necrosis resulting from depletion of natural anticoagulant).

Drs. Hayward and Kelton: Hamilton Health Sciences Corporation, McMaster Campus, 1200 Main Street West, Hamilton, Ontario L8N 3Z5, Canada.

Ms. Johnston and Ms. Russett: Hamilton Civic Hospitals Research Centre, 711 Concession, Hamilton, Ontario L8V 1C3, Canada.