Introduction

Heparin‑induced thrombocytopenia (HIT) is an adverse drug reaction mediated by platelet‑activating immunoglobulin G (IgG) antibodies that target complexes of platelet factor 4 (PF4) bound to heparin^1,2 and confer significant thrombotic risk.^3-5

The immunobiology of HIT is characterized by an immunizing heparin exposure followed by a relatively quick, and generally transient, immune response. Pathologic IgG antibodies, often with concomitant nonpathologic IgA and IgM formation, targeting the PF4/heparin complex are detected by immunoassays at a median of 4 days after the first heparin exposure.⁶ This immune response contrasts with the classical adaptive immunity paradigm in which an early IgM‑predominant response precedes a long‑lasting IgG‑predominant response after class switching. In typical‑onset HIT, the platelet count declines around day 5 to 10 following heparin exposure.^7,8 As discussed in detail below, the pathologic antibody is transient and wanes in detectability in a stereotyped temporal fashion.^6,7 Notably in HIT, there is generally no anamnestic response, meaning that later‑term (>100 days) re‑exposure to heparin may cause neither recurrence of antibody development nor clinical HIT.^5-8

Heparins are frequently used anticoagulants in hospital, dialysis, and operative settings. Nearly one‑third of hospitalized patients will receive heparin.⁹ Though the overall incidence of HIT in hospitalized patients is 0.2%, the incidence rate is reported to be 3‑fold higher in surgical populations.¹⁰ The incidence of HIT approximates 5% after receipt of unfractionated heparin (UFH) for postorthopedic surgery thromboprophylaxis^11,12 and 2% after cardiac surgery with cardiopulmonary bypass (CPB) with intraoperative and postoperative UFH.^13-17 The incidence is dependent on duration and type of heparin used, with longer durations of exposure and UFH (vs low‑molecular‑weight heparin) conferring a higher risk of HIT.^7,10,12

HIT complicated by thrombosis (HITT) occurs in about 40% of HIT cases.^3,4,18-20 Thrombosis has a venous predominance but can be arterial or microvascular, and carries substantial risk of gangrene, limb amputation, and death.^3,4,20,21 HIT may also be complicated by major bleeding due to thrombocytopenia and nonheparin anticoagulant use.^7,21-23

Diagnosis and phases of heparin‑induced thrombocytopenia

When there is intermediate or high suspicion of HIT by the 4Ts score,^24,25 further evaluation by immunoassay is recommended.²⁶ There are different anti‑PF4 / heparin immunoassays available, including polyspecific or IgG‑specific enzyme‑linked immunosorbent assays (ELISA), polyspecific or IgG‑specific chemiluminescent immunoassays, latex agglutination immunoassays, and lateral flow immunoassays, which may have different diagnostic accuracy dependent on test characteristics, thresholds, and antibody specificities but generally have excellent negative predictive value.^27,28 The optical density (OD), though an arbitrary and laboratory‑specific unit, further subclassifies the positivity of the ELISA into low, intermediate, and high thresholds to optimize diagnostic sensitivity and specificity.^27,29-31 The median interval between heparin exposure and positive immunoassay (seroconversion or seropositivity) is approximately 4 to 6 days in typical‑onset HIT,^6-8 with nearly 90% of positive ELISAs demonstrating increased OD between days 4 to 14 when tested serially after heparin exposure.⁶ The OD of immunoassays is typically maximally positive between days 10 and 12 after heparin exposure and generally wanes afterward.⁶ Seroconversion uniformly precedes the platelet count decline.⁸ In individuals with serologically‑confirmed HIT, the median time to a negative test for an immunoassay was 85 days with a 95% CI of 64 to 124 days.⁷

Following a positive immunoassay, the second‑line use of a washed platelet functional assay improves diagnostic accuracy. These functional assays, including the serotonin‑release assay (SRA) and heparin‑induced platelet‑activation (HIPA) test, are more specific than immunoassays for pathologic antibodies and are generally batch‑performed in specialized coagulation laboratory settings.²⁸ Functional assays become negative more quickly than immunoassays at a median of 50 days with a 95% CI of 32 to 64 days.⁷

Based on the timeline of heparin exposure, thrombocytopenia, and the seroconversion and seroreversion of diagnostic assays, HIT follows a stereotyped temporal sequence,^7,32 which is referred to as the phases of HIT (Table 1).³³ Though we present time periods in which this sequence generally occurs, the timeline exhibits substantial variability. Therefore, HIT laboratory testing should be used to identify the phase of HIT in an individual patient and inform the potential risk of heparin re‑exposure for an invasive procedure rather than basing management solely on time from heparin exposure.^6-8

Exceptions to these stereotyped phases include rapid‑onset HIT, which represents a rapid fall in platelet count due to prior immunizing heparin exposure and continued presence of circulating pathologic HIT antibodies,^7,32 and autoimmune HIT in which pathologic heparin‑independent antibodies can arise spontaneously without proximate heparin exposure and generally last for weeks, usually in a postsurgical or postinfectious setting.^34,35 Autoimmune HIT syndromes fall outside the scope of the present review.

Cardiovascular surgery

In acute HIT, a patient will demonstrate thrombocytopenia, a positive immunoassay, and a positive functional assay. In subacute HIT A, thrombocytopenia has resolved, but the immunoassay and functional assay remain positive (Table 1). Acute HIT is the phase in which the majority of thromboses occur.^4,21 Anticoagulation with a nonheparin anticoagulant is strongly recommended in this setting.²⁶ The use of therapeutic‑intensity dosing of nonheparin anticoagulation is recommended over prophylactic‑intensity, as it appears to reduce the risk of new thromboembolic complications.³⁶ Choice of anticoagulant is influenced by availability, need for monitoring of anticoagulant effect, and administration route; patient‑specific factors such as clinical status, kidney and liver function, and risk of bleeding; and operative factors such as need for intraoperative anticoagulation, cardiopulmonary bypass (CPB), and postoperative thromboprophylaxis.

Use of heparin in acute HIT risks thromboembolic events and anaphylactoid reactions. However, UFH has long been a choice intraoperative anticoagulant for patients undergoing cardiac surgery due to its ability to be monitored at the point of care, titrated, and reversed; low cost; and familiarity to surgeons, pharmacists, and perfusionists. Due to the risk of HIT recurrence, American Society of Hematology (ASH) clinical guidelines recommend delaying cardiovascular surgery until the patient has subacute HIT B or remote HIT,²⁶ if feasible. If surgery cannot be delayed, there are no randomized data nor consensus protocols to guide the appropriate approach in acute HIT or subacute HIT A, but cohort studies and case series have reported multiple effective methods of facilitating urgent cardiovascular surgery, including: 1) intraoperative bivalirudin, 2) intraoperative heparin with pre- or intraoperative plasma exchange (PEX), or 3) intraoperative heparin with a potent antiplatelet agent (Table 2). The second and third approaches may also include use of intravenous immunoglobulin (IVIG).

Intraoperative bivalirudin has been studied prospectively in acute HIT during CPB in the CHOOSE‑ON trial³⁷ with the primary outcome of in‑hospital procedural success (absence of death, myocardial infarction, repeat operation, or stroke) achieved in 94% of the 49 evaluable patients, though notably this trial excluded patients with severe renal dysfunction and those requiring surgery on more than 1 cardiac valve. Protocols for the use of direct thrombin inhibitors (DTIs) such as bivalirudin or hirudin for both on‑pump and off‑pump cardiovascular surgeries have been published.^38-40 While the subgroup of patients with glomerular filtration rate (GFR) of 30–60 ml/min/1.73 m² had numerically similar outcomes to the remainder of the population in the CHOOSE‑ON trial, the renal elimination of bivalirudin may result in higher bleeding risk in those with GFR below 30 ml/min/1.73 m² or on hemodialysis, though this may be mitigated by intraoperative hemofiltration.^41-43 In a retrospective comparison of heparin and DTIs as intraoperative anticoagulants during CPB, 13 patients treated with a DTI within 3 months of HIT diagnosis had similar rates of 30‑day mortality and thrombosis as 59 heparin‑treated cases. Patients treated with a DTI appeared to have lower baseline cardiac operative risk and underwent lower risk operations (no cardiac transplants and only 1 aneurysm repair), revealing likely selection bias, but did have lower rates of moderate or severe hemorrhage than heparin‑treated patients after adjustment for thrombocytopenia and time on CPB.⁴⁴ Outside of patients with HIT, a randomized multicenter trial comparing heparin with protamine reversal to bivalirudin in 101 patients undergoing cardiac surgery with CPB demonstrated no difference in the primary end point of in‑hospital procedural success (absence of death, myocardial infarction, stroke, or repeat revascularization) and similar rates of blood loss with no incident thrombosis in the oxygenator, arterial line, or cardioplegia line.⁴⁵

A second approach to intraoperative heparin use in acute HIT or subacute HIT A is perioperative PEX with donor plasma as the replacement fluid. A cohort of 24 patients who received intraoperative PEX prior to intraoperative heparin use for CPB reported 3 (12.5%) cases of thromboembolic events, 2 of which appeared to be HIT‑related due to concomitant platelet count declines, and 3 non‑HIT‑related deaths. Of the 11 patients who underwent both pre- and post‑PEX HIT ELISA testing, the median OD decreased from 1.99 to 0.34, though the time at which the post‑PEX ELISA was sent is unclear.⁴⁶ Another series of 11 patients showed that a single PEX treatment reduced antibody titers by 50%–84% and two‑thirds had ELISA results in the normal range. There was no HIT recrudescence.⁴⁷ In a case report of a patient with subacute HIT A undergoing coronary artery bypass grafting, the patient underwent serial PEX until SRA negativity but with ELISA remaining positive, thus mimicking transition to subacute HIT B. The patient was then treated with intraoperative UFH followed by postoperative fondaparinux with no thrombotic complications reported.⁴⁸ IVIG has been successfully coupled with this approach with the goal of binding platelet FcγRIIa receptors to inhibit HIT antibody‑mediated platelet activation that may last through the initial postoperative period.^49-51

Finally, a third approach involves the use of potent antiplatelet inhibition to facilitate intraoperative heparin exposure. In a patient with subacute HIT A in whom urgent cardiac surgery for atrial thrombectomy was required on day 16 after heparin exposure, the combination of perioperative IVIG and cangrelor, a P2Y₁₂ receptor antagonist, was administered with the cangrelor bolus preceding systemic heparinization by 10 minutes. Serial immunoassays did not appreciably change postoperatively. SRA showed that 77% serotonin release was reduced to 49% after IVIG, then to 11% after cangrelor administration, and then increased to 37% after completion of CPB and thereafter normalized on day 37 after first heparin exposure.⁵² Iloprost and tirofiban have also been used.^53,54 There are no randomized or comparative data to recommend one approach over another. Our guidance is to follow prior institutional and surgical experience. If heparin is used (with either PEX or an antiplatelet agent), it should be strictly limited to the intraoperative setting. If anticoagulation is required before or after surgery, a nonheparin anticoagulant should be used. Postoperative platelet count monitoring is recommended to be performed every 2 or 3 days from day 4 to 14 after intraoperative heparin exposure to monitor for the development of recurrent HIT.⁵⁵

As many patients have experienced thrombosis and require anticoagulation in acute HIT and subacute HIT A, inferior vena cava (IVC) filters may be considered as a means of temporarily reducing the need for therapeutic anticoagulation to facilitate operations. However, placement of an IVC filter is contraindicated as HIT has a predilection for thrombosis at sites of localized endothelial injury,^56,57 which has also been demonstrated for central venous catheters.⁵⁸ Regardless, IVC filter placement does not obviate the need for anticoagulation given the risk of deep vein thrombosis, arterial thromboembolism, or microvascular complications and yields only a small decrement in risk of pulmonary embolism (PE) in those who can tolerate anticoagulation,⁵⁹ and thus their placement is discouraged in this setting.²⁶

To minimize bleeding risk, many invasive procedures have a platelet threshold. The platelet nadir is reached during the acute phase of HIT and generally falls to the level of moderate thrombocytopenia (50–100 × 10⁹/l).²¹ Platelet transfusion has been associated with higher rates of arterial thrombosis and mortality in HIT after adjustment for clinical severity and acuity in a national inpatient database study,⁶⁰ though other retrospective reports have demonstrated no increase in thrombotic events attributable to platelet transfusion.^21,61 Noting the limited evidence, platelet transfusion is not recommended unless there is active bleeding or high risk of bleeding.²⁶

Subacute HIT B and remote HIT are both characterized by resolution of thrombocytopenia, with subacute HIT B having normalization of the functional assay and remote HIT having normalization of both the functional assay and immunoassay (Table 1). The overall body of experience with repeat exposure to heparin in the intraoperative setting suggests a low risk of HIT recurrence in the subacute HIT B or remote HIT phases, though published data are limited to cohorts and case series.

A number of series on the use of heparin in surgical settings in these phases of HIT have been published. In a series of 17 individuals with subacute HIT B (9 patients) or remote HIT (8 patients), intraoperative heparin in cardiac or vascular surgery induced seroconversion of the immunoassay in 11 patients (65%) and SRA in 8 (47%). Timing to seroconversion is similar to typical‑onset HIT on the first presentation, and there was only 1 episode of recurrent HITT, which appeared to be autoimmune HIT given the finding of heparin‑independent platelet activation on SRA.^5,62 A brief course of heparin for cardiac or vascular surgery was also reported in 4 patients with prior diagnoses of HIT after the loss of heparin‑dependent antibodies with no recurrent thrombocytopenia or thrombosis. One of these patients with HIT initially diagnosed 2.5 years earlier developed a positive immunoassay on day 6 and functional assay on day 8 but did not experience recurrent thrombocytopenia or thrombosis.⁷ In a series of 3 patients with subacute HIT B (positive anti‑PF4 / heparin IgG, negative HIPA) who underwent heart transplantation with intraoperative heparin during CPB, none had thromboembolic complications though 2 had an elevation in the IgG‑specific ELISA OD with continued HIPA negativity.⁶³ In cases where emergent cardiovascular surgery is needed and immunoassay or functional assay status cannot be determined prior to surgery, an understanding of the natural history of HIT antibodies may guide consideration for UFH re‑exposure. At 3 months after an episode of HIT, there is less than 5% probability of a positive SRA, and the risk of continued SRA positivity is negligible at 12 months.^5,7,17

Percutaneous cardiac intervention

Bivalirudin is also the preferred approach to patients with acute HIT or subacute HIT A who require percutaneous cardiac intervention (Table 2).²⁶ Outside of HIT, bivalirudin has similar outcomes to heparin in percutaneous coronary intervention (PCI). In a meta‑analysis of trials comparing bivalirudin to heparin plus glycoprotein IIb/IIIa inhibitors in patients undergoing PCI, anticoagulation with bivalirudin yielded no difference in major adverse cardiovascular events (MACE) and resulted in significantly lower risk of major bleeding.⁶⁴ In the randomized BRAVO‑3 trial of 802 patients undergoing transcatheter aortic valve replacement, bivalirudin was noninferior to UFH with respect to the primary outcomes of major bleeding and MACE.⁶⁵

Data for the use of bivalirudin in acute HIT or subacute HIT A for PCI are more limited than those for cardiovascular surgery. In a prospective study of 50 patients with HIT or suspected HIT undergoing PCI, 1 bleeding event and 1 death were reported with no thromboses.⁶⁶ In the BRAVO‑3 trial, the subgroup of patients with HIT had similar bleeding complications with bivalirudin and heparin.⁶⁵ Based on these data, the European Society of Cardiology recommends bivalirudin for PCI in patients with HIT (class I recommendation).⁶⁷

Where bivalirudin is not available, argatroban can be considered for procedural anticoagulation. In a prospective, multi‑center evaluation of argatroban during PCI using activated clotting time to assess anticoagulation, 91 patients who underwent 112 separate coronary interventions (21 repeat PCI) had a satisfactory procedural outcome in 95% of cases, adequate procedural anticoagulation in 98% of cases, and major bleeding in 1.1% of cases. SRA or HIPA was positive in 38% of patients, but outcomes were not reported by HIT antibody status.⁶⁸ The use of danaparoid for PCI has also been reported, though outcomes were not reported by HIT status.^69-71 Fondaparinux is not recommended for primary PCI.⁶⁷

Data regarding PCI in subacute HIT B or remote HIT are sparse. The ASH guidelines suggest the use of bivalirudin over UFH in this setting or use of heparin if alternative anticoagulants are unavailable or clinical experience with these agents is lacking (Table 2).²⁶ As with cardiac surgery, if heparin is used for PCI in a patient with subacute HIT B or remote HIT, it should be limited to the procedure only and scrupulously avoided before and afterward. In such cases, postoperative monitoring of the platelet count through the timeline of typical‑onset HIT of 5 to 10 days after heparin re‑exposure is recommended.⁵