Towards a Conceptual Framework for Diagnosis of Heparin-Induced Thrombocytopenia (HIT).

HIT can be regarded as a “clinicopathologic” syndrome whereby the diagnosis is based upon a compatible clinical picture and the presence of platelet-activating, anti-platelet factor 4 (PF4)/heparin antibodies of IgG class. However, both components of this conceptual framework are problematic. First, there are many potential non-HIT explanations for thrombocytopenia and/or thrombosis in hospitalized patients receiving heparin (“clinical confounders”); and second, prospective studies show that numerous heparin-treated patients develop non-pathogenic anti-PF4/heparin antibodies of no clinical significance (“antibody confounders”). A third issue is that the most specific laboratory assay for pathogenic HIT antibodies (platelet serotonin release assay [SRA]) is performed in only a few reference centers, whereas less specific commercial anti-PF4/polyanion enzyme-immunoassays (EIAs) are widely available. Recently, a clinical scoring system (“4 T’s”) that classifies patients into low-, intermediate- and high-risk groups (LR, IR, and HR, respectively) has been shown to have predictive value in estimating the likelihood of a patient having clinically-significant HIT antibodies. Applying these various concepts, we sought to estimate the potential for HIT “over-diagnosis” inferred from a comparison of “conservative” and “liberal” conceptual frameworks for HIT. The conservative definition required an IR or HR clinical score, as well as that strict laboratory criteria be met for defining pathogenic antibodies (positive SRA [>50% serotonin release] and positive EIG-IgG [>0.45 units]). In comparison, a liberal definition considered HIT in the setting of any positive commercial anti-PF4/polyanion-EIA result (>0.40 units) in any patient evaluated for possible HIT irrespective of clinical score. Our data base consisted of 100 consecutive patients evaluated for possible clinical HIT over a 16-month period in which the “4 T’s” clinical scoring system was applied prospectively. Patients underwent systematic serologic assessment by SRA, EIA-IgG, and EIA-GTI. We found that 16 patients (HR=8, IR=8) met the clinical and serologic criteria for clinical HIT using the conservative framework. In contrast, using the liberal framework, there were 32 patients who would have been considered to have clinical HIT (HR=8, IR=12, LR=12). Thus, twice as many patients (32 vs 16) would have been “diagnosed” as HIT from the liberal perspective. The EIA-GTI absorbance values (in units) for the 16 patients identified using the conservative framework (median=2.39; IQR=2.09–2.70; range=1.46–2.90) were significantly greater (p<0.001 by Mann-Whitney u test, 2-sided) than in the additional 16 patients identified using the liberal framework (median=0.89; IQR=0.54–1.13; range=0.42–2.13). In summary, there is considerable potential for over-diagnosis of HIT (by approximately 100%) depending upon one’s conceptual framework (and available assay) for defining HIT. However, in a center relying upon the EIA-GTI for diagnosis, the designation of a patient as having clinical HIT by integrating the clinical score (IR or HR status required) with a strong positive result in the EIA-GTI (e.g., >1.20 units in this data set) yields a population of patients that overlaps considerably with that identified using a more conservative approach utilizing a more specific diagnostic assay. Thus, by integrating information from the clinical score together with the magnitude of a positive test result, accurate diagnosis of HIT is achievable even when very different assays are employed.