Plasma Exchange for Heparin-Induced Thrombocytopenia: Is There Enough Evidence?

Abstract

The perioperative management of patients who have a diagnosis of heparin-induced thrombocytopenia (HIT) with antibodies to heparin platelet factor 4 (anti-HPF4) and require cardiac surgery with cardiopulmonary bypass (CPB) has challenged clinicians for many years. Heparin has been administered to millions of patients since the first use of CPB to provide predictable and reliable anticoagulation, and to prevent blood from clotting in the extracorporeal circuit. There are numerous advantages of heparin for CPB over currently available alternative anticoagulants, including: 1) familiarity among practitioners; 2) ease of administration; 3) validated anticoagulant monitoring; 4) wide-ranging anticoagulant actions, including platelet inhibition; 5) availability of a reliable reversal agent; and 6) low cost. Although off-label use of direct thrombin inhibitors (DTIs) was initially proposed as a potential solution for patients with anti-HPF4 requiring urgent cardiac surgery with CPB, use of a potent, irreversible anticoagulant likely increases the risk of life-threatening bleeding. There may also be increased risk of clotting, because there is limited experience in monitoring anticoagulation and in guiding appropriate dosing of DTIs during CPB. The bleeding risk may be amplified in certain patients, such as those having repeat cardiac surgery, complex surgeries, ventricular assist device insertions, and heart transplants. Such patients are typically at high risk for excessive bleeding based on both surgical factors and comorbidities such as hepatic and renal dysfunction, and they are often the patients who have anti-HPF4 when they present for surgery, having been exposed to heparin within the preceding 100 days.1 The concern of excessive bleeding with the use of DTIs was initially illustrated in the case report by McDonald et al.2 who described life-threatening (i.e., at 4 L/h) and unresponsive bleeding after a patient received r-hirudin anticoagulation for CPB. Even the use of the shorter half-life DTI, bivalirudin, in patients at low risk for bleeding can be associated with substantial bleeding in a subset of susceptible patients. For example, 6 of 98 patients enrolled in the Evolution-On Study who received bivalirudin had blood loss that exceeded 2.5 L in the first 24 postoperative hours compared with 58 patients who received unfractionated heparin and did not display this degree of blood loss.3 However, these findings have not been consistently observed as illustrated in a recent analysis of 115 patients who received bivalirudin anticoagulation for CPB and who did not develop excessive bleeding.4 We have also observed thrombosis in the venous reservoir despite adequate blood bivalirudin levels, which has prompted us to minimize hemostatic system activation by avoiding retransfusion of cardiotomy blood when we use a DTI for CPB. Nevertheless, until a reversal agent is commercially available for any current or emerging alternative anticoagulant to heparin, potential solutions need to be investigated to provide clinicians with alternatives to prevent development of life-threatening bleeding or clotting within the CPB circuit, which may occur with current alternatives, especially in patients who are at high risk for excessive bleeding. The article in this issue of Anesthesia & Analgesia by Welsby et al.5 is noteworthy in that it provides additional data regarding an approach that may be helpful for the management of patients with anti-HPF4 and who are at increased risk for perioperative bleeding. These investigators expand our limited knowledge of the potential use of plasmapheresis as a means to remove the antibodies, as described previously in a case report.6 Plasmapheresis was originally performed experimentally in 1660 by Dr. Richard Lower on dogs. Subsequently, procedures were performed manually (i.e., manual withdrawal of whole blood with concurrent administration of replacement solutions) in humans in France in 1902 and in the United States in 1914. It was not until 1960 that Solomon and Fahey used plasmapheresis therapeutically and it evolved to an automated process. Although in theory any antibody-mediated illness should clinically respond to a lowering of antibody levels via plasma exchange therapy, this concept is not always the case. A clinical response sometimes depends on a number of variables such as the extent of organ damage as related to the degree of deposition of antibodies at the time of diagnosis (e.g., Goodpasture disease). The lack of response is also likely when the levels of pathologic antibodies within blood do not correlate with the levels in the affected target organ such as the central nervous system (i.e., lowering blood levels does not necessarily result in lower central nervous system levels). Nevertheless, plasmapheresis is an established and potentially lifesaving medical therapy for certain disease states, such as thrombotic thrombocytopenic purpura, myasthenia gravis, and for monoclonal and polyclonal gammopathies such as Waldenström macroglobulinemia. More recently, therapeutic plasmapheresis is being utilized to treat life-threatening rejection of solid organs mediated by donor-specific antibodies to ABO or human leukocyte antigen (HLA) proteins within target organs, especially in the settings of heart and lung transplantation. Although plasmapheresis is generally well tolerated and is rarely associated with catastrophic complications, such as air embolus, anaphylaxis, and infection, it may be associated with mild reactions related to citrate or allergens within either the replacement solutions used (e.g., albumin or plasma) or the extracorporeal plasma exchange system (e.g., ethylene oxide antibodies). Potentially fatal acute complications related to blood product transfusion may occur when plasma is used as the replacement solution, including transfusion-related acute lung injury and anaphylaxis. Bleeding may be either precipitated or aggravated in susceptible patients because coagulation factors are typically reduced by 70%–80% whereas platelet count decreases variably by 10–60 × 103/μL with each exchange procedure. The effective early use of plasmapheresis for the management of HIT has been described by several investigators.7–10 Plasmapheresis has also been used as a rescue therapy in patients with HIT who are refractory to routine therapy enabling mechanical reduction of antibody titers11 or as a bridge for pharmacological immune suppression. Although there has been only 1 previous report of plasmapheresis to acutely reduce pathologic antibodies in a patient with anti-HPF4 undergoing cardiac surgery,6 the hypothesis that this intervention might be effective in preventing HIT with thrombosis (HITT) is biologically plausible. Lowering the level of pathologic antibodies should decrease the risk of thrombotic complications after heparin exposure in patients with anti-HPF4 based on 2 general concepts. First, a direct relationship has been noted between anti-HPF4 concentrations measured by either enzyme-linked immunosorbent assay (ELISA) absorbance values or by percentage release of radioactive serotonin via the serotonin release assay and the propensity to develop thrombotic complications. Second, other antibody-mediated diseases that improve with plasmapheresis, such as thrombotic thrombocytopenic purpura and myasthenia gravis, generally start to respond when antibody levels are still detectable but reduced by 60%–80%. In their retrospective institutional review, Welsby et al. describe 11 patients with anti-HPF4 who were managed with plasma exchange during surgery and who received unfractionated heparin for CPB-related anticoagulation. The investigators found that their approach seemed to be effective with respect to a lack of perioperative complications related to HIT when their patients were reexposed to unfractionated heparin, albeit for a short period when the heparin was restricted to the CPB period. The authors also provide the readership with a reasonable assessment of the limitations of their report in their Discussion section. Nevertheless, there are still several unanswered questions regarding their analysis and the potential usefulness of intraoperative plasmapheresis for the management of patients with HIT or HITT. Strikingly, only 2 of their 11 patients had platelet counts <100 × 103/μL at the time of their surgery, and although there were no obvious thrombotic complications, 8 of the 11 patients had postoperative thrombocytopenia with platelet counts <100 × 103/μL. Ten of eleven patients displayed a postoperative decrease in platelet count, and, although thrombocytopenia is common after CPB, it is impossible to exclude ongoing HIT as a contributing factor in this cohort. It would have been extremely useful to have measured the antibody titers during the postoperative interval to examine this issue. It is intriguing that the authors observed a 50%–84% reduction in antibody ELISA absorbance values with the majority (i.e., 8 of 11) of the patients having absorbance values below the lower limits of normal after a single 1.3 volume plasma exchange procedure as illustrated in their Figure 1. This result is quite impressive because a typical 1.5 volume exchange transiently reduces most protein levels by 78%, whereas five 1.5 plasma volume procedures are generally required to reduce antibody levels by 90%, as a result of redistribution of the antibody from the extracellular space because 55% of immunoglobulin (Ig)G antibodies reside in the extracellular space. The authors did not note the time period for their postprocedure ELISA measurements, which would help elucidate a plausible explanation for their findings of substantially reduced ELISA absorbance values. If measurements were obtained in the initial postoperative period, a substantial reduction in anti-HPF4 values would be expected based on the impact of CPB-related hemodilution and/or volume replacement in the setting of intraoperative bleeding, which would result in further elimination or reduction of these antibodies. Other important unresolved questions with this report involve the postoperative platelet count profiles of these patients and the management this series of patients received. In their Figure 2, the authors illustrate that most patients in their series had a prolonged platelet recovery period after surgery (i.e., 7–10 days). It is uncertain whether this was related to reemergence of antibody titers or to ongoing and accelerated platelet consumption in the first week related to endothelialization of the implanted left ventricular assist devices. Although the authors also mention that a subset of patients received postoperative plasmapheresis on the basis of either a suspicion of HIT or for prevention of graft rejection related to anti-HLA antibody profiles, they neglect to detail how many patients received this therapy and how many procedures were required. This is relevant because the lack of HIT-related thrombotic complications may have been secondary to the therapeutic effects of further plasma exchange therapy in the postoperative period. This leads us to the question of what should be done in the postoperative period to monitor and treat these patients. Although a subset of patients in this series (n = 4) was treated postoperatively with anticoagulants that are not implicated in the generation of anti-HPF4, for reasons other than suspicion of HIT, it would have been helpful to follow anti-HPF4 ELISA absorbance and/or serotonin release assay values serially to determine whether ongoing anticoagulant therapy would have been indicated based on theoretical risk of HITT. Perhaps one can also make the argument that anti-HPF4 antibody levels should be monitored postoperatively to identify whether the brief heparin exposure resulted in an increased immune response and on that basis to identify patients who would theoretically benefit from further plasmapheresis procedures as well. Although we are left with some unanswered questions, the investigators should be congratulated for examining this treatment paradigm. What we now need is to perform further, more controlled analyses and perhaps even small, randomized trials to examine the clinical utility of this approach and to extend at least the monitoring, and perhaps the treatment period, into the postoperative period. More sophisticated monitoring methods, such as the serotonin release assay, should be included to confirm the diagnosis of HIT and better define potential thrombotic risk. Controlled studies might shed more light on the potential benefit of plasmapheresis in the perioperative management of patients with anti-HPF4, HIT, and HITT. In the meantime, we are left with suggestive evidence that this approach may be helpful in reducing the risk of life-threatening bleeding and perhaps thrombosis in this high-risk patient population. We should also consider further investigation of other alternative approaches involving either use of short-acting platelet inhibitors such as prostacyclin analogues12–14 or short-acting IIb/IIIa receptor inhibitors such as tirofiban.15,16 GLOSSARY Heparin-induced thrombocytopenia (HIT): a clinical syndrome that involves a decrease in platelet count, usually exceeding 30% from baseline, which occurs after exposure to unfractionated or low-molecular-weight heparin and is not attributable to another cause of thrombocytopenia. Heparin-induced thrombocytopenia and thrombosis (HITT): HIT that is complicated by arterial or venous thrombosis. Anti-HPF4 antibodies: antibodies to the heparin platelet factor 4 complex, which are usually immunoglobulin (Ig)G class antibodies. ELISA for anti-HPF4 antibodies: enzyme-linked immunosorbent assay screening test for all (i.e., IgG, IgM, IgA, etc.) or subtype specific (i.e., IgG) anti-HPF4 antibodies, including those that cause HIT (i.e., IgG) and those that do not. Serotonin release assay (SRA): serotonin release assay is a functional test that determines whether a patient's plasma cause release of radio-labeled serotonin from donor platelets when they are exposed to unfractionated heparin. The SRA may be used to confirm (i.e., in the setting of a high pretest probability and a negative ELISA result) or exclude (i.e., in the setting of a low pretest probability with a positive ELISA result) the diagnosis of HIT.