Severe autoimmune heparin‐induced thrombocytopenia postcardiac surgery: Implications for subsequent cardiac surgery

A 64-year-old New Zealand European male was transferred to Waikato Hospital (tertiary-level facility) with recent non-ST elevation myocardial infarction and severe mitral valve regurgitation secondary to severe triple-vessel coronary artery atherosclerosis. Comorbidities included diabetes mellitus, atrial fibrillation, hypertension, and gout. Coronary artery bypass grafting and bioprosthetic mitral valve replacement were performed; two platelet transfusions were given intraoperatively. Postoperative complications included profound vasoplegia requiring intra-aortic balloon pump (IABP), multiple vasopressors, and acute renal failure. Patients who undergo complex cardiac surgery with combined coronary revascularization and valve replacement are at increased risk for postoperative cardiac failure requiring life support measures such as IABP and vasopressor therapy.1, 2 From the hematologist's perspective, such patients are at increased risk for thrombocytopenia, coagulation abnormalities, bleeding/anemia, and need for blood product replacement.3 During the first 120 h (5 days) postsurgery, the platelet count fell from 251 (preoperative value) to 87 × 109/L on postoperative day (POD) 4, with the downward slope accelerating during the first 36 h following IABP placement (Figure 1; until end of POD4). Associated laboratory abnormalities included: elevated lactic acid (peak, 4.7 mmol/L [reference range, 0.6–2.4]), progressive rise in alanine transaminase (ALT) to peak value of 792 U/L (reference range, 0–55) on POD3, rise in the international normalized ratio (INR) to 2.1 (reference range, 0.8–1.2), elevated activated partial thromboplastin time (APTT) values (e.g., 43 s [reference range, 25–38] at time of peak INR), and first detection of nucleated red blood cells (nRBCs) on POD2. Fibrinogen rose to elevated levels beginning on POD2 (peak on POD4, 6.5 g/L [reference range, 1.5–4.0]). The patient's overall clinical picture improved: by POD4, the IABP was removed, and by POD5, vasopressin and dobutamine were stopped (low-dose norepinephrine was continued). The initial platelet count fall to approximately 100 × 109/L by POD3 reflects the postcardiac surgery state, where ~50% declines by POD2–POD3 are expected.4 In this patient, intraoperative platelet transfusions were given, so clearance of transfused platelets exacerbated the early platelet count fall. The use of IABP is associated with further platelet count decrease.5 Overall, the patient's platelet count changes are expected for this clinical situation. The elevated INR indicates either transient liver dysfunction (“shock liver”), disseminated intravascular coagulation (DIC), or both. However, this patient exhibited elevated fibrinogen levels that remained steady from PODs 2 to 5 (peak INR on POD4), arguing somewhat against a major effect of DIC (however, there is an entity known as “high-fibrinogen DIC” in some patients who develop ischemic limb losses, where a more severe degree of thrombocytopenia would be expected6). By the end of POD4, the IABP had been removed, lactic acidemia had resolved, ALT levels had peaked and were beginning to decline, and the INR had nearly normalized; in combination, these clinical and laboratory events suggest that the patient's presumptive thrombocytopenia of critical illness should also stabilize and improve. However, over the subsequent 120-h period (Figure 1; POD5 to end of POD9), the platelet count decline accelerated, reaching a nadir value of 9 × 109/L (POD9); on POD6, a cold purple left foot, with pulses, was noted (not the limb that had had the IABP). Limb ischemia in critical illness can indicate localized problems (e.g., deep vein thrombosis, limb artery thrombosis) or systemic events such as DIC associated with circulatory shock (cardiogenic, septic).7 Whereas absent pulses suggest large artery occlusion, limb ischemia with detectable pulses can indicate microvascular thrombosis. Indeed, bilateral microvascular thrombosis involving lower (and sometimes upper) limbs—manifesting as “symmetrical peripheral gangrene”—is associated with the triad of acute DIC, shock, and proximate acute ischemic hepatitis (shock liver); here, procoagulant/anticoagulant imbalance results from DIC with shock liver predisposing to severe depletion of hepatically synthesized natural anticoagulants, protein C, and antithrombin.6, 7 Although the discussant (Theodore E. Warkentin) routinely measures protein C and antithrombin activity levels in this clinical setting, this is infrequently performed in most medical communities. On POD6, when limb ischemia was first noted, the platelet count measured 31 × 109/L, the INR was 1.5, and the fibrinogen was only 0.8 g/L. No test for fibrin-specific marker, such as d-dimer, was performed. The peripheral blood film showed red cell fragments; moreover, prominent normoblastemia (circulating nRBCs) was evident from POD2 to POD11. Pan-culture yielded no evidence of microbial infection. This patient's laboratory picture is consistent with DIC. Postcardiac surgery patients develop prominent elevated fibrinogen (acute phase reactant) levels by POD3 (~5–6 g/L)8; thus, a fibrinogen <1.0 g/L is striking. Also consistent with DIC is the presence of red cell fragments (schistocytes). Normoblastemia is typically seen in critically ill patients with progressive limb microthrombosis,6, 7 likely reflecting tissue hypoxia (circulatory shock) and DIC with red cell fragmentation and an associated bone marrow response.9 Table 1 lists potential explanations for DIC in critical illness. No evidence of sepsis was found in the patient. Further, his shock state had resolved (normal lactate levels) when thrombocytopenia became severe. An uncommon explanation for DIC to consider is severe heparin-induced thrombocytopenia (HIT), particularly the variant, “autoimmune heparin-induced thrombocytopenia (aHIT),” caused by highly pathological aHIT antibodies that activate platelets even in the absence of heparin10, 11; overt DIC with hypofibrinogenemia has been reported in such patients.12, 13 Although other (non-heparin) drugs, such as antibiotics, can cause severe thrombocytopenia,14 this entity mimics acute immune thrombocytopenia (isolated severe thrombocytopenia without coagulopathy). Careful review of medications, particularly heparin exposure, is indicated. Medications included preoperative dabigatran, simvastatin, allopurinol, felodipine, glycerol trinitrate, aspirin, clopidogrel, and metformin; postoperatively, the patient received multiple inotropes, perindopril, and, from POD4, empiric meropenem treatment. Table 2 summarizes heparin exposure, which included three injections of enoxaparin (100 mg per dose) given 7–9 days preoperatively, and a single 5000-unit unfractionated heparin (UFH) injection given at coronary angiography 8 days pre-surgery. Intraoperative UFH totaled 85 000 units. Only a single 40 mg dose of enoxaparin was given in the early postoperative period (POD2); importantly, no heparin product was given beyond POD2. The intra-/perioperative exposure to heparin products warrants careful consideration for a diagnosis of HIT. A recent study of 129 subjects with serologically confirmed HIT postcardiac surgery15 showed that 75% of patients develop HIT between POD5 and POD10, inclusive; in the minority (10%) that developed “early” HIT (prior to POD5), all had received heparin during the 10 days prior to cardiac surgery, with the timing of preoperative heparin explaining occurrence of early postoperative HIT. However, in this case, the immediate preoperative platelet count (251 × 109/L) was higher than values obtained earlier in the hospitalization, arguing somewhat against preoperative heparin exposure being responsible for early postoperative HIT. Further, as discussed, the early postoperative platelet count declines appeared consistent with the patient's critical illness. A useful way to approach a patient with possible postcardiac surgery HIT is to plot the serial platelet counts and to look closely for a platelet count fall occurring during the POD5–POD10 window characteristic of HIT.15 In this patient, the dramatic platelet count decline observed, from 87 (shortly before POD5) to 11 × 109/L (POD7), with repeat platelet count of 10 × 109/L later that same day, a nearly 90% platelet count fall, occurring rapidly and without ongoing heparin exposure, is consistent with severe aHIT superimposed in a patient with early postoperative critical illness.16 Applying the 4Ts score17 likely underestimates the likelihood of HIT in this patient. The timing of onset of the postoperative platelet count fall is at best unclear, if not too early (0–1 point); the platelet count nadir is unusually low (0 points); there is no clear thrombosis (limb ischemia points to “possible” thrombosis, 1 point); and the patient is critically ill, although improving (0–1 point). Overall, the 4Ts score (maximum, 3 points) could be construed as “low probability.”17 Yet, a “gestalt” assessment, by an experienced clinician, would point out that the accelerating platelet count decline in the POD5–POD10 window, occurring several days after the last exposure to heparin, together with overt DIC, in a patient who otherwise seems to be improving, points strongly to a potential diagnosis of severe aHIT. A HIT screen, performed late on POD5 using the particle gel immunoassay (PaGIA; Bio-Rad Laboratories, Auckland, NZ), was positive. However, confirmatory HIT testing with a platelet activation assay was not performed (the serotonin release assay [SRA] is not available in New Zealand). The PaGIA is a rapid HIT test with high diagnostic sensitivity that is not available in the United States, but which has been used in New Zealand and elsewhere.18 (Recently, the PaGIA was discontinued worldwide.) Like other PF4-dependent immunoassays, a positive result is not specific for HIT, especially postcardiac surgery, where nonclinically significant anti-platelet factor 4 (PF4)/heparin antibodies are often formed.19 PaGIA diagnostic specificity is high (>99%) if patient serum diluted 1/32 yields a positive result18 (retesting with diluted serum was not performed in this case). A similar situation holds for other PF4-dependent immunoassays—for example, strong positive results by enzyme immunoassay or chemiluminescence assay indicate a high likelihood of HIT.20 Treatment for HIT is warranted in this clinical situation. Treatment for HIT was initiated with the direct thrombin inhibitor (DTI), bivalirudin, at a renally adjusted dose of 80 μg/kg/h (for estimated glomerular filtration rate (eGFR) 53 mL/min) and monitored by APTT (target therapeutic range, 80–100 s). However, platelet recovery was slow (Figure 2); a normal platelet count was not reached until POD32. The overall clinical picture of severe thrombocytopenia occurring during the POD5–POD10 day period, accompanied by DIC, microvascular limb ischemia, and delayed platelet count recovery, is strongly suggestive of aHIT.10 This severe manifestation of HIT is gaining increasing attention, for several reasons. First, such patients have highly pathological aHIT antibodies that cause strong platelet activation in vitro, using the SRA, even in the absence of heparin, that is, heparin-independent platelet-activating antibodies.10, 21 Second, these highly pathogenic antibodies recognize antigen sites on PF4 similar to those caused by vaccine-induced immune thrombotic thrombocytopenia (VITT) antibodies,22 a severe prothrombotic anti-PF4 disorder.21 Third, management differs from classic HIT, as high-dose IVIG is recommended to interrupt aHIT antibody-induced platelet activation.23 Fourth, these patients are at greater risk for anticoagulant treatment failure—including progression of limb ischemia to necrosis, including with DTI treatment, perhaps due to “APTT confounding” (inadequate DTI dosing due to difficulties in determining “therapeutic” APTT values) as well as DTI inhibition of protein C activation by thrombin.24 Accordingly, the discussant favors anticoagulation with factor Xa inhibitors (e.g., fondaparinux, danaparoid [not available in the United States], rivaroxaban, or apixaban), agents that do not require APTT monitoring.24 The patient's limb ischemia resolved within 48 h, he exhibited good clinical recovery and was discharged on POD27, on warfarin (target INR, 2.0–3.0) and aspirin 100 mg po daily; the maximum INR measured during warfarin thromboprophylaxis was 2.0. At 3-month postdischarge, warfarin was changed to dabigatran 150 mg per os twice daily. HIT treatment guidelines suggest postponing initiation of warfarin anticoagulation until platelet count recovery; this is because warfarin, administered during acute HIT, is a risk factor for microthrombosis, particularly with supratherapeutic INR levels (>4.0).24 However, in this patient, no adverse effect of warfarin was seen. Two years later, the patient represented to hospital with prosthetic valve endocarditis associated with Enterococcus faecalis. Medical management failed and a plan was made to proceed with redo sternotomy and mitral valve replacement. Preoperatively, the platelet count was normal, and a screening test for HIT antibodies was negative (lateral flow immunoassay [STAGO ANZ STic Expert, Forest Hill, Australia]). Because of concerns regarding the potential for HIT recurrence, bivalirudin rather than heparin was used for intraoperative anticoagulation. There are differing opinions on how to handle this situation of repeat heart surgery in a patient with a remote history (>3 months prior) of HIT, with negative testing for HIT antibodies. The American Society of Hematology (ASH) guidelines suggest using heparin for intraoperative anticoagulation, since HIT recurrence (if it occurs) will take at least 5 days, at which point non-heparin anticoagulation can be given. However, there is a small chance—estimated at less than 3%—of HIT recurrence, as a manifestation of aHIT, the entity I believe the patient had 2 years earlier. Indeed, a well-documented example of recurrent aHIT with repeat heparin exposure was reported in a patient whose two postcardiac surgery episodes of HIT—separated by 20 years—represented two distinct episodes of aHIT.25 Intraoperative bivalirudin was initiated at 1 mg/kg at knife-to-skin and then infused at 2.5 mg/kg/h until just before cardiac bypass was discontinued. Citrated-dextrose was used to prime the cell saver. Bivalirudin was monitored with activated clotting time (ACT), aiming for a target range of 2.5 times baseline. Redo cardiac surgery was technically challenging due to extensive adhesions. Initial low-level bleeding during careful exposure of the surgical field was exacerbated by the anticoagulant effects of bivalirudin; approximately 2500 mL of blood was returned from the cell saver. The infected mitral valve was successfully explanted and replaced. Due to a patent left internal mammary artery graft, surgery proceeded with moderate hypothermia (30°C) for myocardial protection. Bivalirudin was stopped just before the cardiac bypass was disconnected. Ongoing bleeding and persistently raised ACT delayed chest closure for 4½ h, with numerous intra-/early postoperative blood products given (4 units red cell concentrates [RCCs], 4 units fresh frozen plasma [FFP], 50 units [~500 mL] cryoprecipitate, 1 platelet transfusion), and postoperative acute kidney injury documented. Approximately 10 h post-chest closure, cardiac tamponade prompted return to the operating room, with the evacuation of large retrosternal hematoma and additional transfusions (10 units RCCs, 4 units FFP, 20 units [~200 mL] cryoprecipitate). Seven RCC units were also given over the subsequent 5 days. Ultimately, the patient made a good recovery. Use of hypothermia may have resulted in delayed bivalirudin clearance and prolonged anticoagulant effect. The challenges of major intra-/early postoperative bleeding associated with bivalirudin anticoagulation highlight why many practitioners choose intraoperative anticoagulation with heparin—despite a history of previous aHIT—in this situation. Fortunately, the patient's bleeding was manageable. This patient illustrates several challenging aspects regarding the diagnosis and management of HIT. First, the patient's clinical course is consistent with a highly atypical clinical presentation consistent with the severe subtype, aHIT. Such patients can develop severe thrombocytopenia, overt DIC, and microvascular thrombosis, with delayed platelet count recovery despite the absence of ongoing heparin administration. Second, the case illustrates how test availability for HIT antibodies varies in different jurisdictions; for example, in New Zealand, it is challenging to obtain a platelet activation test such as the SRA. Even in the United States—where the SRA can be performed—it is uncommon for the test to be performed both in the absence and presence of heparin; yet, definitive laboratory diagnosis of aHIT requires a functional (platelet activation) test performed both with and without heparin, as strong serum-induced serotonin release in the absence of heparin is the pathological hallmark of aHIT.10 Third, this case illustrates the treatment dilemma that arises when a patient who is believed to have a diagnosis of severe aHIT requires repeat cardiac surgery. In theory, such a patient can receive heparin safely for the cardiac surgery itself (given the absence of HIT antibodies at the time of surgery) with vigilance in the postoperative period to evaluate for potential aHIT occurrence during the characteristic POD5 to POD10 period; at this time, if aHIT occurs, a non-heparin anticoagulant plus high-dose IVIG would be appropriate. Alternatively, one can perform the surgery with a non-heparin anticoagulation strategy, such as with bivalirudin. In this case, the second approach was adopted. Although major postoperative bleeding occurred, the ultimate outcome was favorable. We thank Ms. Jo-Ann I. Sheppard for preparing the two figures. Theodore E. Warkentin has received lecture honoraria from Instrumentation Laboratory (Werfen); has provided consulting services to Ergomed, Instrumentation Laboratory (Werfen), Octapharma, Paradigm Pharmaceuticals, and Veralox; has received research funding from Instrumentation Laboratory (Werfen), and has provided expert witness testimony relating to HIT and non-HIT thrombocytopenia. Khalid Al-Azri, Kate Goldstone, Julia Phillips, Jack Bhana, and Nishith Patel declare no relevant conflicts of interest. The patient provided written permission for this report (institutional approval is not required for case reports in our medical community). All data generated or analyzed during this study are included in this published article.