Laboratory tests for heparin-induced thrombocytopenia: a multicenter study.

A multicenter clinical trial of the thrombin inhibitor argatroban (Novastan; Texas Biotechnology, Houston, TX; Smith-Kline Beecham Pharmaceuticals, Philadelphia, PA) was recently conducted in patients with heparin-induced thrombocytopenia (HIT) and HIT that had progressed to thrombosis (HITTS). In patients defined by the inclusion/exclusion criteria, the utility of three diagnostic HIT assays was investigated: the platelet aggregation assay, the serotonin release assay (SRA), and the enzyme-linked immunosorbent assay (ELISA) for the antibody to the heparin-platelet factor 4 (H-PF4) complex. Confirmation was made in 26%, 55%, and 64% of the patients, respectively (n = 199 patients; 512 to 606 samples; P < .001 platelet aggregation assay v SRA v ELISA). Patients who progressed to HITTS (n = 98) were more often confirmed than were HIT patients without associated thrombosis (n = 101) (P < .05). Confirmation by platelet aggregation assay and SRA results generally was associated with a higher antibody titer. However, a minimum critical titer could not be identified, because all patterns of positive and negative results by the platelet aggregation assay, SRA, and ELISA were observed, and clinically ill patients had a wide range of antibody titers. Over a 30-day period, the percentage of positive responses did not change. Although multiple testing over several days enhanced the chance of confirmation, this difference was not significant. Combined results of the three assays enhanced the positive response to 83% of the total population (P < .005). These data demonstrate that there is no direct correlation between the positive response of these assays, and that clinically positive HIT patients can be missed by all three assays. With these limitations, the combination of platelet aggregation assay, SRA, and ELISA testing with multiple samples offers the best chance of confirming a positive HIT patient. Caution is advised, however, in interpreting all assay results, as no assay is optimal.