Comparison of platelet antibody screen, crossmatching and HLA antibody testing in patients refractory to platelet transfusions

Abstract

Patients undergoing recurrent platelet transfusions can become refractory to these transfusions. Platelet antibody screens (Immucor), platelet crossmatching assays (Immucor), and HLA antibody testing are commonly used to test these patients. The relative effectiveness of these tests has not been determined. A higher incidence of strongly positive screen results that did not predict crossmatch results was anecdotally noted. Therefore, the results of the platelet antibody screens and crossmatches were systematically compiled over a 12-year period from 2010 to 2021. Of note, the Immucor Capture-P Ready Screen (platelet antibody) had a recall in March 2013 after which the performance of the test appears to have changed. The positivity rate of the platelet antibody screen increased over the course of the study, and this was statistically significant when analyzing year as a continuous variable and when grouping years by four-year periods (2010-13,2014-17,2018-21). In contrast, platelet crossmatch reactivity decreased slightly throughout this period. During the 2018-21 period, HLA antibody testing was commonly performed and correlated well with the crossmatch testing but not with the screen. These results suggest that the drastic increase in positivity we observed in the platelet antibody screen over this period is due to increased analytic sensitivity (with possible reduced specificity) of the screen and not a change in our patient population. Based on these results, the platelet antibody screen has little clinical utility and directly performing platelet crossmatching or HLA antibody testing is recommended for patients suspected to be refractory to platelet transfusions due to alloimmune-mediated factors.

Keywords: Platelet antibody screeen; Platelet refractory; Platelet transfusion; Platelets.

Figure 1:. Increase in positivity of platelet antibody screens over time.

The percentage of positive wells in the first platelet antibody screen (Screen) and first crossmatch tests for each patient is shown as indicated. The numbers above the data points indicate the total number of tests performed each year. Reactivity rates of the platelet antibody screen increased significantly over time (OR=1.40, 95% CI 1.35–1.46, p < 0.001), while rates of the crossmatch testing slightly decreased (OR=0.96, 95% CI 0.93–0.99, p=0.014). Statistical analysis was performed as described in the Materials and Methods Section.

Figure 2:. Reactivity rates of platelet antibody screens and crossmatch testing before and after the platelet antibody screen recall.

The percentage of reactive wells in all patients’ first platelet antibody screen (Screen) and crossmatch testing is grouped from 2010–2013 (before the recall) and from 2014–2017 and 2018–2021 (after the recall). Numbers above bars indicate total number of tests performed in each 4-year period. The reactivity rate of platelet antibody screens increased significantly after the recall (2010–13 vs 2014–17: OR=3.60, 95% CI=2.80–4.65, p<0.001; 2010–13 vs 2018–21: OR=14.6, 95% CI=10.5–20.8, p<0.001). The reactivity rate of platelet antibody crossmatch decreased significantly after the recall (2010–13 vs 2014–17: OR=0.52, 95% CI=0.38–0.70, p<0.001; 2010–13 vs 2018–21: OR=0.70, 95% CI=0.53–0.91, p=0.008). Statistical analysis was performed as described in the Materials and Methods Section.

Figure 3:. Positivity Rates of Platelet Antibody screen, HLA testing, and Crossmatch Testing from 2018–2021.

The mean positivity percentages of platelet antibody screens (screens), HLA antibody tests (reported as panel-reactive antibody (PRA)), and platelet crossmatch tests from 2018–2021 performed within 21 days of each other. The error bars represent the standard error of the mean. The average positivity of these screens was significantly higher than both the average PRA (p=0.0012) and the average crossmatch positivity (p=0.0056), while the average PRA and crossmatch positivity were not statistically different (p=0.5167). Statistical analysis was performed as described in the Materials and Methods Section.