A Laboratory Investigation in Response to Clinical Concern Regarding Heparin-PF4 ELISA Accuracy. (11th September 2019)

• Record Type: Journal Article
• Authors: Carll, Timothy; Wool, Geoffrey
• Is Part Of: American journal of clinical pathology. Volume 152(2019)Supplement 1
• Page Start: S22
• Page End: S22
• Publication Date: 2019-09-11
• Subjects: Diagnosis, Laboratory -- Periodicals; Pathology -- Periodicals; 616.07
• DOI: 10.1093/ajcp/aqz112.042
• Languages: English
• ISSNs: 0002-9173

Abstract

Abstract: Heparin-induced thrombocytopenia (HIT) is an immune-mediated disease that confers increased risk of thromboembolism to some patients receiving heparin. Usual evaluation of HIT entails a combination of clinical assessment ("4T" score), screening assay for anti-PF4 antibodies, and confirmatory functional assays, most commonly 14 C-serotonin release assays (SRAs). At this academic medical center, clinician perception of increasing positivity on the screening ELISA raised concern for testing false positivity or a locoregional increase in HIT incidence. In response, we reviewed all anti-PF4 IgG/A/M ELISA testing for UChicago Medicine patients in a 3.5-year period in order to assess the accuracy of this assay in comparison to gold-standard SRA testing. ELISA results were categorized as negative (<0.400 OD), weakly positive (≥0.400 but <1.000 OD), or strongly positive (≥1.000 OD). During the same interval, all instances of confirmatory testing in a heparin-induced serotonin release system were compiled and scored by result as negative (<20% release difference with low-dose [0.1 U/mL] heparin and no apparent inhibition by high-dose [100 U/mL] heparin), borderline (<20% release difference with low-dose heparin, with 8% or more inhibition by high-dose heparin), and positive (>20% release with low-dose heparin, together with inhibition of release [<20%] with high-dose heparin). Both ELISA and SRA testing were stratified by month of order to highlight temporal trends. In total, 1, 657 instances of anti-PF4 ELISA testing were studied, of which 298 were weakly positive and 166 were strongly positive (464 total abnormal ELISA). There were 348 instances of SRA testing, of which 50 were positive and 7 were borderline (57 total abnormal SRA). Of note, there were no instances wherein a negative ELISA test was followed by a borderline or positive SRA test. Over the most recent 4-month interval, a slight increase in the percentage of ELISA positivity was observed. A concordant increase in the number of positive SRA tests was also observed. There was no evidence of an increased rate of discrepancy between the two testing platforms, suggesting that this change does represent a locoregional increase in HIT cases. We took this opportunity to evaluate our ELISA testing sensitivity, specificity, and positive and negative predictive values and found them to be similar to previously published metrics. Given SRA testing as a gold-standard metric, this suggests a local sensitivity and negative predictive value of 100%. As expected, a strongly positive PF4 result was found to be a stronger predictor of a borderline or positive SRA result (39.8% PPV) than was a weakly positive PF4 result (4.5% PPV). As a note of caution, however, one patient had a weakly positive PF4 ELISA result and 100% release on SRA testing.