Utility of D-dimers in the diagnostic work-up of heparin-induced thrombocytopenia (HIT)

Introduction The most widely used approach for diagnostic of HIT is based on the combination of the 4T score and immunoassays (IA) detecting anti-PF4/heparin antibodies. In our center, we employ a Bayesian diagnostic algorithm incorporating the 4T score and the magnitude of two sequential rapid IAs. Because HIT is characterized by an activation of the coagulation system, and based on our previous experience, we assessed whether the quantitative result of D-dimers could improve the diagnostic work-up.

Method We are currently conducting a prospective validation of our in-house diagnostic algorithm (01.2017 – 10.2022). A confirmatory functional HIPA (heparin-induced platelet activation) test is performed for either a high 4T score and/or any not negative IA result(s). Among this cohort (n = 282), a D-dimer analysis was performed using INNOVANCE D-Dimer of the CS-5100 System (Siemens). The diagnostic performance of D-dimers was evaluated by ROC analysis, allowing us to determine the AUC, the optimal cut-off value, likelihood ratios (LR) of result intervals, and the 100% negative predictive value (NPV).

Results Among the 282 analysed samples for which HIPA results were available, HIT could be proven by a positive HIPA in 142 cases (50.4%). The AUC of the ROC curve was 0.711 and the optimal cut-off was identified at a D-dimer value of 3746 ng/ml (sensitivity 82.4%, specificity 50.7%). The 100% negative predictive cut-off value (NPV) for a positive HIPA was at 1’000 ng/ml. Of note, 14/140 (10%) of non-HIT samples were below the 100% NPV cut-off value. The D-dimer interval from 1’000 ng/ml to 3746 ng/ml had a LR of 0.432; the one from 3746 ng/ml to 5’000 ng/ml a LR of 1.808; from 5’000 ng/ml to 30’000 ng/ml of 1.121; and the D-dimer interval >30’000 had a LR of 6.080. The 100% positive predictive cut-off value (PPV) could not be determined due to the lack of specificity of D-dimers.

Conclusion According to our preliminary date, a D-dimer value below 1’000 ng/ml could be used to exclude HIT. LR of higher result intervals could be useful to modify pretest probability for HIT. However, high D-dimers do not predict a positive HIPA and cannot be used to predict HIT.

Publikationsverlauf

Artikel online veröffentlicht:
20. Februar 2023