Evaluation of a rapid and automated heparin‐induced thrombocytopenia immunoassay

Abstract Introduction Heparin‐induced thrombocytopenia (HIT) is a potentially life‐threatening adverse reaction of heparin. Laboratory evaluation of HIT is often not available within a reasonable time. We evaluated the HemosIL® HIT‐Ab(PF4‐H) (Instrumentation Laboratory), a rapid, on‐demand, fully automated, latex immunoturbidimetric assay (LIA). Materials and methods Following determination of the LIA's reference interval and cutoff values, a multicenter study was conducted between March 2013 and June 2015. Plasma samples of HIT‐suspected patients (n = 632) were collected and evaluated by LIA on the ACL TOP® Family systems (Instrumentation Laboratory), enzyme‐linked immunosorbent assays (EIA), and serotonin release assay (SRA). Patient characteristics, medical conditions, comorbidities, laboratory results, and medications were collected via medical chart review. The pretest clinical probability of HIT was also calculated for each patient. Results Based on the 95% reference interval for healthy donors and HIT‐negative patients, a LIA value ≥1.0 U/mL was interpreted positive. The overall agreement of LIA versus EIA and SRA results were 90% (95% CI 88%‐92%) and 79% (95% CI 75%‐82%), respectively. The negative predictive value for LIA and EIA was comparable (87%) with SRA. The positive and negative percent agreements with the clinical probability were 89% (95% CI 69%‐97%) and 86% (95% CI 83%‐89%), respectively, with a negative predictive value of 99.6% (95% CI 98%‐100%). Discussion Overall, the LIA results were comparable to those of EIA and SRA. This fully automated assay with a remarkable short analytical turnaround time of <20 minutes can be performed on‐demand, which would greatly facilitate more prompt management of HIT.

complex-mediated adverse reaction. Due to the widespread use of heparin, HIT is considered one of the most common adverse drug reactions, and the most frequent form of immune-mediated druginduced thrombocytopenia. If unrecognized, HIT may be associated with significant morbidity and mortality. Unfortunately, since thrombocytopenia is very common in various medical conditions, a diagnosis of HIT may be missed.
Clinically, HIT may be suspected in patients who develop, otherwise unexplained, significant drop in platelet count (≥50%) and/ or development of thrombosis within 5-10 days after the start of unfractionated heparin (UFH) or low molecular weight heparin (LMWH). Although about 8% of patients receiving heparin are at risk for developing HIT antibodies, only 1%-5% will develop thrombocytopenia, and approximately one third of these may develop arterial and/or venous thrombosis. 3,4 The risk of HIT depends on the type of heparin, length of exposure, and patient risk factors. For example, the incidence is higher among patients receiving UFH (almost 10fold) versus LMWH 6 and in patients undergoing major versus minor surgeries. 7 Heparin forms complexes with platelet factor-4 (PF4), a circulating plasma protein that is mainly secreted by platelets, in vivo. The immunogenicity of these glycoprotein structures results in the rapid generation of IgG antibodies (between day 5 and 14), 8 which, in turn, bind to the PF4-heparin complex, forming PF4-heparin-IgG immune complexes. These immune complexes crosslink FcγRIIa receptors on the membranes of platelets 9 and monocytes (FcγRI) 10,11 causing their activation that, along with possible alterations of endothelial cells, 12 promotes thrombin generation.
Typically, the platelet count in HIT starts to fall 5-10 days 13 after the initiation of heparin therapy; however, in case of a previous exposure (within 90 days), thrombocytopenia may evolve within minutes to hours of re-exposure to heparin due to preformed antibodies, resulting in a more rapid-onset form of HIT. 13,14 In either case, heparin-PF4 antibodies may persist in the plasma for an additional 2-3 months. 4 The diagnosis of HIT is primarily based on clinical judgment, but since HIT is considered a clinical-pathological syndrome, this diagnosis should be confirmed by appropriate laboratory evaluation.
Laboratory confirmation of HIT can be achieved by platelet activation (functional) assays, or using immunoassays that can directly detect heparin-PF4 antibodies. Immunoassays are more commonly performed and are technically less demanding. These assays generally have high sensitivity (80%-100%) but suffer from low specificity due to the detection of antibodies that are not platelet-activating, and therefore do not elicit HIT (false positives). 15 Enzyme-linked immunosorbent assays (EIA) are most widely used to support the diagnosis of HIT; however, this test platform typically requires sample batching, which may delay confirmation of the diagnosis of HIT, and the EIA may be unavailable in laboratories with low test volumes.
Platelet activation assays for HIT (ie, the serotonin release assay [SRA] and heparin-induced platelet activation [HIPA]) test measure platelet activity in the presence of the patient's serum and multiple concentrations of heparin. These assays are more specific than EIAs for clinically relevant HIT antibodies.
However, these assays are technically demanding and time-consuming. They are therefore restricted to specialized laboratories and are usually utilized after positive immunoassay results have been obtained, further delaying confirmation or exclusion of the diagnosis.
Since laboratory test results are often not available within a reasonable time (usually hours to days), it is essential to discontinue all heparin exposure in strongly suspected HIT patients, including heparin-coated catheters and heparin flushes, and initiate alternative anticoagulation. If bridging to warfarin has already been started in acute HIT patient, warfarin therapy should be postponed pending thrombocytopenia recovery (platelet count >150 × 10 9 /L) 9 as warfarin predisposes to microvascular thrombosis. 16,17 We evaluated the HemosIL ® HIT-Ab (PF4-H) (Instrumentation Laboratory, Bedford, MA), a rapid, on-demand, fully automated, latex immunoturbidimetric assay, as an aid in the diagnosis of HIT and provides results more quickly than existing tests. The assay detects HIT antibodies through a reaction mechanism that competitively inhibits agglutination of HIT-like monoclonal antibody-bearing particles. 18

| Study design
This is a cohort study to investigate the diagnostic accuracy of a newly developed automated, on-demand HIT assay HemosIL ® HIT-Ab (PF4-H) (Instrumentation Laboratory). In addition, we aimed to demonstrate the substantial equivalence of this latex immunoturbidimetric assay (LIA) to a reference method (Serotonin Release Assay) and to a predicate method (Stago Asserachrom HPIA-IgGAM ELISA assay, Diagnostica Stago, Asnieres, France). Patient's medical record information including demographic characteristics (sex, age, race, and ethnicity), medical conditions, comorbidities, medications, provenance (cardiovascular surgery, orthopedic surgery, intensive care units, etc), type of heparin (UFH, LMWH), and the original clinical HIT antibody test result was collected via medical chart review.

| Patient population
The pretest clinical probability of HIT was calculated for each patient in the study using the 4T Score HIT Assessment Point System. 19 Additionally, the overall clinical probability was calculated using the 2013 American Society of Hematology (ASH) guideline that was adopted in part from the 9th edition of the American College of Chest Physicians Evidence-Based Clinical Practice Guidelines, 20 a diagnostic algorithm that classifies samples as "HIT Likely" or "HIT Unlikely" based on the "4Ts" score, EIA, and SRA results ( Figure 1).

| Sample handling and processing
As part of routine clinical care, a freshly collected (within 4 hours) sodium citrated sample was obtained from each HIT-suspected patients at time of HIT testing at each center. Leftover plasma samples that were no longer needed for clinical testing were processed, and three aliquots (500 L each) were generated for each sample and kept frozen at a minimum of −70°C until testing. Samples were then rapidly thawed at 37°C, or per institution's process for thawing, centrifuged at 1500 xg to remove precipitates and homogenized adequately prior to testing. Hemolyzed, icteric, lipemic, partially clotted, and/or turbid samples were excluded as interferences of this assay were determined by the manufacturer at the following levels: free hemoglobin >500 mg/dL, bilirubin >19 mg/dL, and triglycerides >375 mg/dL. Samples were de-identified, and a unique study ID was assigned to each enrolled study specimen prior to study-related testing. All samples were tested following only a single freeze-thaw cycle when possible.
Samples were tested by trained clinical laboratory technologists using the LIA on the ACL TOP Family system (Instrumentation Laboratory) and by EIA (Asserachrom ® HPIA assay; predicate assay; n = 632; see below). Tests were performed in duplicate according to the manufacturer's package insert for the predicate assay. An aliquot of each sample was also shipped on dry ice to the Florida Hospital Center for Thrombosis Research (Winter Park, FL) for SRA testing (n = 537).

| Serotonin release assay
The SRA was performed as previously described 21 to detect heparin-dependent platelet-activating anti-PF4/heparin antibodies.
Briefly, following the collection of citrated whole blood, platelet-rich plasma was prepared via differential centrifugation. Platelets were then washed in calcium-and albumin-free Tyrode's buffer and radioactively labeled with 14C-serotonin and incubated with heat-inactivated patient plasma or control plasma. Samples were tested in

| HemosIL ® HIT-Ab (PF4-H) assay
As previously described, 18  Reference interval in heparin-exposed and hit-suspected patients (HIT negative) For this group, samples were collected from patients who were exposed to heparin and suspected of having HIT but had tested negative by either of the commercially available EIAs (Asserachrom HPIA assay or and GTI-PF4 Enhanced EIA, Immucor GTI Diagnostics, Inc, Waukesha, WI). As above, citrated whole blood samples (4.5 mL) were collected from 122 patients. Samples were then processed and analyzed by LIA utilizing ACL TOP Family systems.

| Receiver operating characteristic curve analysis
A method comparison study with the SRA was used to confirm the LIA cutoff value by Receiver operating characteristic (ROC). Sixtythree frozen plasma samples from HIT-suspected patients with low, moderate, and high 4Ts score (31 were confirmed positive and 32 were confirmed negative by SRA) were analyzed by LIA utilizing ACL TOP Family systems.

| Statistical analysis
Reference interval studies data were analyzed using Microsoft Excel Partitioned reference intervals were shown to be unnecessary using the Student t method with two-sided 95% confidence interval (CI) to compare the central locations of the distribution of LIA results across gender. The reference interval of the heparin-exposed and HIT-suspected patients (HIT Negative) was calculated according to the CLSI EP28-A3c guidelines. ROC curve analysis was performed using CLSI EP24-A2 bias estimation guidelines. The data were also analyzed using Microsoft Excel and the data analysis add-on package Analyse-It. The ninety-five percent (95%) CI for total agreement of LIA vs SRA was calculated using Wilson's score confidence interval.
The clinical studies data were analyzed using Microsoft Excel

| Determination of the HemosIL HIT-Ab (PF4-H) assay reference interval
The reference interval was calculated using citrated plasma samples from normal donors (n = 131) and HIT-suspected patients F I G U R E 2 The HemosIL HIT-Ab (PF4-H) assay Reference Interval. The cutoff was determined by testing blood samples of healthy donors (n-131) that were tested negative for HIT by EIA; and HITsuspected patients (n = 122) that were exposed to heparin but were confirmed negative for HIT antibody by both of the commercially available EIAs (Asserachrom HPIA assay and GTI-PF4 Enhanced EIA, Immucor GTI Diagnostics, Inc, Waukesha, WI). The 95% reference interval was 0-0.7 for the healthy donors and 0-0.9 for the HIT-suspected patients

U/mL Cut-off
Healthy donors HIT suspected Heparin exposed

| Receiver operating characteristic curve analysis
The optimal cutoff value confirmed by ROC analysis was 1.0 U/mL.
The ROC curve showed the pattern of an informative assay, and the area under the curve (AUC) was calculated to be 0.95 (95% CI
Comparing LIA to SRA after sorting the patients into three different groups (Supplementary Table S1) according to their pretest clinical probability of HIT (ie, "4T" score, low: 1-3; moderate: 4-5; or high: 6-8) revealed better PPA and PPV in the high 4Ts score group (78.6% and 55%, respectively). These results also corresponded to the Asserachrom HPIA EIA assay when compared to the SRA (85.7% and 57.1%, respectively; supplementary Table S2). and low pretest probability of HIT. 18,23 Timing is critical in the diagnosis of HIT and may affect patient management and clinical decisions in HIT-suspected patients. Even in laboratories that perform the EIA daily, the use of batch testing can still result in turnaround times up to 30 hours. 24 The advantage of an on-demand, automated assay is that test results can be provided in a more timely fashion, even in a routine clinical laboratory setting.

| D ISCUSS I ON
The "4T"score was introduced in 2003 25 and is a commonly used means of predicting the pretest likelihood of clinical HIT using a scale of 0-8 points. If the pretest probability score falls between 0 and 3, HIT is clinically unlikely; a score of 4-5 indicates

| Conclusions
In summary, the LIA test for HIT antibodies showed comparable results to the commercially available EIA and SRA. This rapid, fully

ACK N OWLED G EM ENTS
This study was funded by Instrumentation Laboratory.

CO N FLI C T O F I NTE R E S T
The authors have no competing interests.