In this issue of Blood, Samuelson Bannow et al describe the results of a multicenter, prospective, blinded study that compared the performance of the serotonin release assay (SRA) against the platelet factor 4 (PF4)-dependent P-selectin expression assay (PEA) for the diagnosis of heparin-induced thrombocytopenia (HIT).1
HIT is a potentially life-threatening, prothrombotic, immune complication of heparin caused by immunoglobulin G antibodies that recognize complexes of PF4 and heparin. Thrombocytopenia and thrombosis in HIT are caused by a subset of anti-PF4/heparin antibodies that elicit cellular activation by binding and cross-linking platelet FcγRIIa.2 Recent studies suggest that antibody binding to FcγRIIa on monocytes3 and on neutrophils4 contributes significantly to thrombosis in HIT.
Central to the diagnosis of HIT is laboratory identification of heparin-dependent, platelet-activating antibodies. In most circumstances, HIT testing begins with an immunoassay to detect the presence of anti-PF4/heparin antibodies. A variety of immunoassays for HIT on various testing platforms are currently available, including solid-phase, particle gel, and latex immunoturbidometric assays. Although these assays all have excellent sensitivity (>98% to 99%), they all suffer from poor positive predictive value (38.7%5 -55.6%)6 and high rates of false positives. Therefore, in circumstances in which clinical evaluation and immunoassay results are not adequate to definitively confirm diagnosis, functional testing is often necessary.
Functional testing for HIT is based on the key principal that pathologic, clinically relevant anti-PF4/heparin antibodies are capable of activating platelets. At this time, the gold standard functional test to confirm HIT is the SRA, although other assays with different platelet activation end points are available. In the SRA, platelets from a normal donor are incubated with 14C. After washing, 14C-loaded platelets are incubated with heat-inactivated patient plasma or serum in the presence of buffer, heparin at a therapeutic dose (0.1-0.3 U/mL), or heparin at an excess dose (100 U/mL). Importantly, exogenous PF4 is not required for the SRA, presumably because washed platelets release an adequate amount of PF4 for immune complex formation. After incubation and centrifugation, 14C-serotonin released into the supernatant is quantified by a β-counter.7 Despite improved sensitivity and specificity (>88% for both) compared with the immunoassays,8 the SRA is technically challenging, requires reactive platelet donors (only one third of normal subjects are suitable donors for HIT testing),7 and uses radioactivity. Because of these challenges, the SRA is not readily available at most institutions and is often limited to reference and commercial laboratories. As a result, clinicians are often left to make management decisions that are based on clinical suspicion and immunoassay results alone as they await SRA results, which may take days.
Because of the SRA’s limitations, in recent years, a functional assay that does not require radioactivity and that may be more easily adapted by laboratories, the PEA, has been developed. The PEA is a platelet activation assay that uses washed platelets pooled from 3 donors. After incubation with PF4 and patient serum, platelet P-selectin expression is measured by flow cytometry and is expressed as a percentage of maximal P-selectin expression (after treatment with thrombin receptor-activating peptide). In contrast to the SRA, the PEA requires exogenous PF4 to be added, and anti-PF4/heparin antibodies are presumed to recognize complexes of PF4 bound to platelet surface glycosaminoglycans. Previous retrospective studies have shown that the PEA has diagnostic accuracy that is similar to, if not higher than, that of the SRA.9,10 However, those studies used archived, annotated patient samples for which serologic testing results were already known, and this may have resulted in selection bias.
In their study, Samuelson Bannow et al performed blinded SRA and PEA testing on 409 consecutive adult inpatients with suspected HIT from 2 study sites. Each patient sample was tested using the same donor platelets for both assays. Following a prespecified diagnostic scheme, clinical information and enzyme-linked immunosorbent assay results were used to determine whether the patient was positive, indeterminate, or negative for HIT, allowing for analysis of sensitivity and performance characteristics for both assays. The authors found that the sensitivity and specificity of the PEA was comparable to that of the SRA (sensitivity 76% vs 82% and specificity 96% vs 97% for the PEA vs SRA, respectively). Negative concordance between the PEA and SRA was 0.974 and positive concordance was 0.692. Examination of the discrepant cases showed that 5 of 10 patients among the PEA-positive/SRA-negative patients likely had HIT, and 3 of 8 patients in the SRA-positive/PEA-negative group likely had HIT.
On the basis of their data, Samuelson Bannow et al support the use of the PEA as a functional test for HIT. In contrast to the SRA, the PEA does not require radioactivity, and therefore, may be more readily adopted by laboratories, which will provide clinicians with more rapid results. The importance of minimizing turnaround time for functional testing is important. As highlighted in the Samuelson Bannow et al study, 49% of patients who were treated with an alternative anticoagulant were later deemed to be negative for HIT when functional test results came back negative. One of these patients suffered a clinically relevant non-major bleed. More rapid turnaround of functional testing would potentially minimize the risk of patients being unnecessarily treated with alternative anticoagulants, which are associated with a higher risk for bleeding.
Although Samuelson Bannow et al have established that the PEA is comparable to the SRA in terms of assay performance, several concerns must be addressed before it can be widely implemented. This assay does not require radioactivity, but it remains technically challenging because it requires washed platelets and expertise with platelet flow cytometry. These requirements may not expand the availability of the assay beyond the testing centers that currently offer the SRA, and therefore the issue of turnaround time may not improve. In the Samuelson Bannow et al study, most PEA tests were performed at a single site that had considerable expertise, so it remains to be seen whether assay performance characteristics will be similar in the hands of other operators. Finally, the findings of discordant results in the PEA and SRA are concerning and require further study to fully define the role of the PEA in confirming (or excluding) the diagnosis of HIT. Nevertheless, development and validation of the PEA addresses a long-standing diagnostic challenge in this field.
Conflict-of-interest disclosure: The author declares no competing financial interests.