A rapid test (STic Expert® ) for the diagnosis of heparin-induced thrombocytopenia.

Abstract

We read with great interest the recently published report (Leroux et al, 2014) on the evaluation of a rapid nano-particle based flow immunoassay (STic Expert® HIT) for the diagnosis of heparin-induced thrombocytopenia (HIT). The diagnosis of HIT is very difficult and relies on clinical and laboratory data. The prognosis depends on the early recognition of the disorder, and as this adverse drug reaction causes great mortality, it is extremely important that initial HIT screening methods have an optimal sensitivity. We would like to report our results from a similar study. We collected citrated plasma samples of 153 patients suspected with HIT between May 2006 and January 2014 (stored at −80°C). Samples were thawed at 37°C in a warm water bath before analysis with STic Expert® HIT (Diagnostica Stago, Asnières sur Seine, France) and enzyme-linked immonosorbent assay (ELISA; Asserachrom® HPIA IgG; Diagnostica Stago), according to the manufacturer's instructions. Results for STic Expert® HIT were read by three readers and confirmed with a flow-cytometric CD62p (p-selectin) functional assay (Tomer, 1997; Tomer et al, 1999; Denys et al, 2008), which identified 20 HIT-positive samples (13·07%). For the ELISA, a cut-off was calculated for each run as advised by the manufacturer. The risk of HIT was estimated by the Warkentin 4T score (Greinacher & Warkentin, 2006) and accordingly classified as low, intermediate and high risk (Table 1). As described by Leroux et al (2014) we also found 2 HIT patients in the low risk group (4T score of 3 for both cases), indicating that HIT cannot be totally excluded by a low 4T score. As opposed to the report by Leroux et al (2014), our HIT cases did not include cardiac surgery patients but a patient receiving heparin therapy for port-a-cath®-related thrombosis and a patient receiving low molecular weight heparin after non-orthopaedic surgery. The sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) for STic Expert® HIT and ELISA were determined using Bayesian analysis (MedCalc® Online diagnostic test calculator, http://www.medcalc.org/calc/diagnostic_test.php). The inter-reader agreement for STic Expert® HIT was calculated on the basis of kappa agreement (spss statistics 22.0; IBM, Armonk, NY, USA and R: a language for statistical computing, R Foundation for Statistical Computing, Vienna, Austria). Asserachrom® HPIA IgG showed an excellent sensitivity of 100·00% [95% confidence interval (CI) 83·01–100·00] and a good specificity of 80·45% [72·68–86·81]. PPV and NPV were 43·48% [28·94–58·89] and 100·00% [96·58–100·00] respectively. The inter-reader reproducibility for STic Expert® HIT (kappa ratio) varied from 0·800 to 0·983 between readers. Overall Fleiss' kappa agreement between the three readers was 0·875. This is lower than that obtained by Leroux et al (2014). Sensitivity was 95·00% [75·05–99·17] for all readers and specificity for different readers varied from 83·46% [76·03–89·33] to 84·21% [76·88–89·95]. In our study, STic Expert® was negative for the 3 readers in 112 cases, but the diagnosis of HIT could only be excluded in 111. The sensitivity for this assay was therefore 95·00% [75·05–99·17] with a NPV of 99·15% [95·35–99·86], which is similar to the 99·6% reported by Leroux et al (2014) who attributed their only false negative result in the plasma sample group to a probable error of the operator by forgetting to add the patient sample to the cassette. Our only false negative result was considered negative by the 3 readers. The sample concerned the earlier mentioned patient with a Warkentin 4T score of 3 (low risk) receiving low molecular weight heparin after non-orthopaedic surgery. The negative result with STic Expert® was confirmed on the defrosted sample by another operator and three different, blinded, readers. In contrast, Asserachrom® HPIA IgG gave a positive result with an optical density of 0·725 (cut-off 0·251). The positive result for HIT was confirmed with a functional CD62p flow-cytometric assay. This is in agreement with another recent study (Vianello et al, 2015) in which the NPV was calculated to be 97% for STic Expert® due to 2 false negative samples from 114 cases. The authors did not specify whether this concerned serum or plasma samples. We therefore conclude that, although a rapid nano-particle-based flow immunoassay, such as STic Expert® HIT, can be very useful for the diagnosis of HIT, caution is appropriate as false negative results cannot be excluded in the case of frozen, citrated plasma samples. The authors wish to thank Michael Luypaert for his technical assistance and all technologists for performing the analyses. LDC performed the analyses, analysed the data and wrote the paper; KDV identified the collection of samples, designed the study, analysed the data, revised the data analysis and paper. There are no conflicts of interest.