Improving specificity in HIT testing

Richard H. Aster

Heparin-induced thrombocytopenia (HIT) affects approximately 5% of patients given unfractionated heparin for 7 days. Although thrombocytopenia itself is rarely serious, a subset of patients experience arterial or venous thrombosis that can be life-threatening. Because continuation of heparin can be catastrophic, early diagnosis of HIT is critical so that heparin can be discontinued and alternative treatment instituted. Algorithms based on the timing and severity of thrombocytopenia, presence or absence of thrombosis, and a lack of other causes for the drop in platelets have been devised to aid clinicians in assessing the likelihood that a particular patient has HIT.1

Antibodies associated with HIT are remarkably specific in nearly all cases for complexes comprising 2 normal body constituents: the highly acidic glycosaminoglycan heparin and the basic chemokine, platelet factor 4 (PF4; CXCL4).2,3 Two types of laboratory tests are available for detection of these antibodies: one uses platelet activation as an end point and the other is based on the reaction of antibodies with immobilized complexes made up of PF4 bound to heparin or one of several other negatively charged macromolecules (PF4 enzyme-linked immunosorbent assay [ELISA]).

The best-studied platelet activation assay measures release of radiolabeled serotonin from target platelets exposed to patient serum and a low concentration of heparin (usually 0.1 units/mL). In patients with clinical findings suggestive of HIT whose serum is positive in the serotonin release assay (SRA), the diagnosis of HIT is highly likely.3 Unfortunately, the SRA is technically demanding and is not widely available except on a referral basis. In contrast, the PF4 ELISA is simple to perform, can be done in a few hours, is highly sensitive and specific for antibody detection, and is widely available in the form of several Food and Drug Administration–approved kits. The diagnosis of HIT is highly unlikely in a patient who tests negative in the PF4 ELISA test.2,3 However, owing in part to its high sensitivity, the PF4 ELISA often detects antibodies that would not be positive in the SRA and may be clinically insignificant.

A clinician caring for a patient suspected of having HIT is unable to wait several days for an SRA test to be performed before prescribing treatment. Therefore, it is important that the PF4 ELISA test provide as much information as possible about the likelihood that a patient has HIT. One way to improve this correlation is to consider the optical density (OD) value of the test result because a patient whose serum produces an OD value of 1.0 or more is much more likely to have HIT that one whose serum is “positive” but in the range 0.4 to 1.0.3,4 A second strategy is to configure the test to detect only antibodies of the immunoglobulin G (IgG) isotype because antibodies of the IgA and IgM classes rarely trigger HIT.3

As first shown by Visentin et al,5 PF4-heparin complexes formed at a heparin:PF4 molar ratio of approximately 1:2 are optimal for HIT antibody detection, and reactions of antibodies with this target are inhibited when high doses of heparin are added. A significant fraction of serum samples from patients suspected of having HIT react with the unmodified complexes, sometimes quite strongly, but fail to inhibit with excess heparin. What these antibodies recognize is uncertain, but they are not likely indicative of HIT. In this issue of Blood, Whitlatch et al describe studies in which they examined the clinical status of 212 patients who tested positive using a PF4 ELISA test that detected IgG, IgA, and IgM antibodies.6 Retrospectively, they then determined the extent to which use of a heparin inhibition “confirmatory” step improved the correlation between a positive PF4 ELISA test result and the clinical diagnosis of HIT. Findings showed that fewer than 10% of patients judged to have HIT on clinical grounds failed the confirmatory test, whereas approximately one-third of those who did not appear to have HIT did so. Using statistical modeling, they derived a nomogram by which the likelihood of a patient having HIT can be estimated (1) using the OD value in PF4 ELISA and (2) by determining whether or not the reaction is inhibited by at least 50% when excess heparin is added. Using this tool, a patient whose serum produces an OD value of 1.0, for example, would have an 80% chance of having HIT if the confirmatory test is positive but only a 20% chance if it is not.

A clinical diagnosis of HIT calls for heparin to be discontinued, usually followed by start of treatment with a direct thrombin inhibitor, and, later, oral anticoagulation. This can mean extra days in the hospital and many thousands of dollars in additional costs. The findings of Whitlatch et al may help to reduce this burden for patients who do not actually have HIT. However, there are several caveats. In the Whitlatch study, approximately 18% of the “positive” PF4 ELISA test results failed to confirm with excess heparin. In contrast, Warkentin et al, using the same PF4 ELISA test kit, found that only 2 of 78 HIT antibodies (3%) failed to confirm with excess heparin.7 The reason for this discrepancy is unclear. However, in a study of 6000 samples tested in PF4 ELISA (R.H.A., B.R. Curtis, unpublished, April 1, 2010), our laboratory found that the percentage of positive tests that failed to inhibit was comparable with the 18% observed by Whitlatch et al. Second, many laboratories now test only for antibodies of the IgG class in the PF4 ELISA. The extent to which this modification influences the number of “positive” test results that fail the confirmatory test with excess heparin needs to be determined.

National Institutes of Health


  • Conflict-of-interest disclosure: The author declares no competing financial interests. ■