Correlation of PF4 ELISA Results and Clinical Likelihood of HIT: Can We Improve on Specificity?.

Abstract

The diagnosis of heparin-induced thrombocytopenia (HIT) is made clinically with the support from the laboratory. While the Serotonin Release Assay (SRA) is considered by some to be the “gold standard” diagnostic test for HIT, this complex assay is not widely available. The PF4 ELISA is readily available, but is positive in many patients (pts) who do not have HIT. Measures suggested to improve the specificity of the PF4 ELISA include detecting only IgG antibodies and regarding only high potency antibodies as confirmatory evidence for HIT. We sought to correlate the clinical findings in 77 pts whose samples were tested in the PF4 ELISA with the serologic results (OD in PF4 ELISA for IgGAM, G, A and M and SRA) to determine if strength of reaction or antibody class is related to the clinical likelihood of HIT or to the results of the SRA. In addition, because occasional sera are positive in PF4 ELISA, but fail to inhibit by at least 50% when excess heparin is added (“reactive” samples), we examined whether this confirmatory (heparin inhibition) step provides additional specificity to the test result. We reviewed clinical information on the pts and assigned each a clinical “4T” score (Lo et al, J Thromb Haemost, 2006) estimating the likelihood of HIT. The IgGAM PF4 ELISA results were categorized as negative (OD < .4), reactive (OD >.4 with < 50% inhibition), weakly positive (OD.4 – 1.0) and strongly positive (OD > 1.0). The analysis showed that the pts with strongly positive results had 4T scores and SRA activity that were significantly higher than those of the other groups (Mean score 5.16 and SRA 61% Vs 2.5 – 3.7 and 8 – 10%, TTEST p=.02 – .00001). The weakly positive group had somewhat higher scores than the negatives and reactives (3.75 Vs 2.5 and 2.6, p = .02 and .04) but did not differ from either of these groups in SRA activity (8.0% Vs 10% and 8%, p = .4). The reactive and negative groups had similar scores and SRA activity despite 6/20 of the reactive samples having ODs >1.0. The fraction of samples with either high clinical scores (≥ 6) or high SRA activity (≥ 75%) did not differ among the negative, reactive, or wk positive groups, but was significantly higher in the strongly positive group (CHITEST, p =.0005) (Tab). Fewer samples were classified as weakly positive or reactive when the IgG only PF4 ELISA was used with no reduction in the number of strong positives (Tab ). Of note, one sample that was strongly positive in the IgGAM assay (OD 1.2) and negative in the IgG only test had a strong IgM antibody (OD 1.1). This pt had a clinical score of 6 and likely had true HIT. A second sample that was weakly positive in both the IgGAM and IgG only assays was strongly positive in the IgA only test (OD 1.4). This pt also had a clinical score of 6. 6 additional samples were weakly or strongly positive in the IgGAM assay without IgG antibody and had IgM with or without IgA antibodies. The clinical scores were 0.5, 2, 2.5, 4, 4.5 and 5, suggesting that clinical HIT was unlikely in 3 and possible in the remaining 3.IgGAM CategoryNo. of PtsNo. with Clin Score ≥ 6No. with SRA > 75%IgG Only CategoryNo. of PtsNo. with Clin Score ≥ 6No. with SRA > 75%Negative Reactive2411Negative44212011Reactive700Weak Pos1420Weak Pos710Strong Pos1999Strong Pos191011These findings confirm previous reports showing that a strongly positive PF4 ELISA test result (using either the IgGAM or the IgG-only assay) correlates well with a positive SRA and with the likelihood of having HIT. However, pts with reactive PF4 ELISA results generally have a negative SRA and a low 4T score and are an exception to this rule. The heparin inhibition step identifies reactive samples found in 1–2% of all pts referred for PF4 ELISA testing and improves the correlation between test results and the likelihood of having HIT. Finally, while the IgG only PF4 ELISA reduces the number of weakly positive or reactive results and preserves nearly all of the clinically relevant strong positives, we have identified at least 2 definite and 3 possible examples of relevant non-IgG HIT antibodies, 4 of which would have been missed using an IgG only test. Since pts have been described who had apparent HIT with thromboembolic complications whose antibodies were solely IgM, testing for IgG antibodies only could have adverse consequences for these exceptional individuals.