Limitations of the particle immunofiltration assay test for diagnosis of heparin‐induced thrombocytopenia

Department of Epidemiology, School of Public Health, University of Alabama at Birmingham, Birmingham, Alabama Department of Pediatrics, Bayero University/Aminu Kano Teaching Hospital, Kano, Nigeria Department of Radiology, Bayero University/Aminu Kano Teaching Hospital, Kano, Nigeria Department of Pediatrics, Hasiya Bayero Childrenʼs Hospital, Kano, Nigeria Murtala Mohammed Specialist Hospital, Kano, Nigeria Department of Pharmacy, Aminu Kano Teaching Hospital, Kano, Nigeria Department of Radiology, Barau-Dikko Teaching Hospital, Kaduna, Nigeria Department of Hematology, Bayero University/Aminu Kano Teaching Hospital, Kano, Nigeria Department of Medicine, Murtala Muhammad Specialist Hospital, Kano, Nigeria Department of Psychiatry, Bayero University/Aminu Kano Teaching Hospital, Kano, Nigeria Rodeghier Consultants, Chicago, Illinois Division of Hematology and Oncology, Department of Pediatrics, Vanderbilt-Meharry of Excellence in Sickle Cell Disease, Vanderbilt University Medical Center, Nashville, Tennessee Department of Pediatrics, Divisions of Pediatric Hematology-Oncology and Clinical Pharmacology and Toxicology, University of Arkansas for Medical Sciences, Little Rock, Arkansas Division of Hematology and Oncology, Department of Medicine Vanderbilt University School of Medicine, Nashville, Tennessee Department of Pediatrics, University College of London, Great Ormond Street Institute of Child Health, London, UK Division of Pediatric Neurology, Department of Pediatrics, Vanderbilt University Medical Center, Nashville, Tennessee Vanderbilt Institute for Global Health, Nashville, Tennessee

results are not usually available on the same day of blood draw. Since HIT is ultimately diagnosed in only a minority of patients investigated per clinical suspicion, 6 and given the need for timely treatment decisions, there is growing interest in rapid immunoassays for HIT. 7,8 The particle immunofiltration assay [PIFA (HealthTEST Heparin/Platelet Factor 4 Antibody Assay; Akers Biosciences, Inc., Thorofare, NJ]), a rapid immunoassay for detection of PF4/heparin antibodies, received clearance by the U.S. Food and Drug Administration (FDA) in 2004. 9,10 In 2016, Sun et al 7 included the PIFA in a systematic review of rapid immunoassays for HIT diagnosis based on one study, 11 which found 100% PIFA sensitivity, albeit with a wide confidence interval (95% CI, 0.05-1.00). The wide CI resulted from only two SRA-positive study patients (as discussed later, these likely were false-positive SRA results).
In contrast, PIFA specificity was only 0.687 (95% CI, 0.586-0.773). Further, these investigators 7 were not able to include in their review an earlier 2007 study we reported, 10 as our results were presented graphically (as ROC curve analyses) without providing the numerical data needed for inclusion in the systematic review. This likely also explains why our joint Hamilton/Greifswald PIFA evaluation (assessing 289 samples, including 25 HIT-positive patients) 10 was not included in a later systematic review of rapid immunoassays by Nagler et al. 8 In the meantime, additional data on the PIFA has become available, [12][13][14][15] including two studies 13,14 presented in abstract form at the recent ASH annual meeting (December 2019). We now report the results of our analysis involving the sensitivity and specificity of the PIFA in all English language studies reported to date, 10-15 along with a recent study evaluating a modified PIFA, the PIFA PLUSS. 16 (The PIFA PLUSS includes a seraSTAT Rapid Blood Cell Separator, allowing for testing of whole blood, rather than serum. 16 ) Prompted by a recent report, 17 we also obtained a report on the results of proficiency testing for the PIFA. Details regarding our systematic review and data synthesis are provided in a supplemental file which includes a PRISMA Flow Diagram ( Figure S1 in Appendix S1) and a QUADAS-2 assessment of study quality (Table S1 in Appendix S1).
We performed three analyses. First, we estimated PIFA sensitivity and specificity for those studies that determined HIT-positive status by washed platelet activation test (SRA or HIPA) as the reference standard. If the study indicated that a particular sample was positive by SRA or HIPA but negative by PF4-dependent EIA, the sample was regarded as HIT-negative. This reduces risk of a false-positive functional assay result, 9 and also avoids potential bias towards too negative PIFA assessment because these sera might also not be recognized by other antigen tests. The 95% CIs for the individual studies were computed based on the method of Wilson, 18 as recommended by Agresti and Coull 19 for small samples. Overall estimates of PIFA sensitivity and specificity were obtained by jointly synthesizing the data from all seven studies using a bivariate random effects model for meta-analysis of diagnostic test data, which accommodates study heterogeneity. 20 Second, for those studies that evaluated samples by both the PIFA and an EIA, 10,11,13,16 we constructed 2 × 2 tables by cross-classifying samples according to the two methods. We then assessed the level of agreement between the two assays using Cohen's kappa statistic along with associated 95% CIs. 21 An overall measure of agreement was then computed by taking a weighted average of the study-specific statistics using weights proportional to the inverse of the variances in order to maximize the precision of the resulting estimate.
Third, we obtained the results of a proficiency testing exercise for PIFA which was conducted from 2011 to 2019 by the External Quality Control for Assays and Tests (ECAT) Foundation. In this program, external laboratories tested two samples; one HIT-positive, the other HITnegative. We determined yearly outcomes of participating laboratories obtaining the expected result of positive or negative for the two samples tested. Figure 1A shows the seven studies (in six reports [10][11][12][13][14][15] ) which evaluated the PIFA against a platelet activation reference standard.
In our evaluation of the Miami study, 11 both SRA-positive patients were classified as HIT-negative based upon negative EIA results (these patients also had low 4Ts scores and were not regarded by the study authors as having had HIT). 11 We therefore also assessed overall PIFA sensitivity and specificity omitting the Miami study (as there were no HIT-positive subjects to judge test sensitivity). We performed another analysis omitting the Brooklyn study (which was reported in abstract form in 2014 and did not give a comparison with an EIA). 12 We also performed an additional analysis omitting the Gainesville study (as this study appeared to be an outlier). 15 All estimated sensitivities were below 0.714, corresponding to values too low for an acceptable screening test; further, no analysis showed an estimated specificity greater than 0.575. Figure 1B shows those studies 10,11,13,16 that permit comparison of PIFA reactivity vs an EIA. None of the five studies yielded CIs demonstrating improvement over chance agreement. Moreover, when pooling the kappa statistic across studies, the overall measure did not suggest agreement beyond chance. Indeed, the overall raw agreement (pooled data) showed only 51.5% agreement. These results contrast with data presented on two FDA websites, 22,23 suggesting assay performance may have changed. We note that poor assay performance can cause problems in patient management, as illustrated by a report 24  The poor performance of the PIFA is clear from Figure 1A providing additional data regarding one of the studies. 11 We also thank Jo-Ann I. Sheppard for preparing Figure 1 and Figure S1, and Ker-Ai Lee for assistance with statistical calculations.

CONFLICT OF INTEREST
T.E.W. has received lecture honoraria from Alexion and Instrumenta-