Cumulative donor-specific antibody threshold predicts platelet transfusion response in HLA-alloimmunized patients

Abstract

Up to a third of patients with hemato-oncologic conditions who have received multiply transfusions develop immune-mediated platelet transfusion refractoriness. Yet factors that influence posttransfusion platelet corrected count increments (CCI) in patients with HLA-alloimmune platelet transfusion refractoriness remain less well elucidated. Recent advances in HLA antibody characterization using fluorescent bead-based platforms enable the study of donor-specific antibody (DSA) avidity (as measured by mean fluorescence intensity [MFI]) and its impact on HLA-alloimmune platelet transfusion refractoriness. In this large retrospective study of 2012 platelet transfusions among 73 HLA-alloimmunized patients, we evaluated the impact of cumulative HLA DSA-MFI alongside other donor, platelet component, and patient characteristics on CCI at 2 and 24 hours after transfusion. As part of a quality improvement initiative, we also developed and tested a computerized algorithm to optimize donor-recipient histocompatibility based on cumulative DSA-MFI and sought other actionable predictors of CCI. In multivariate analyses, cumulative HLA DSA-MFI of ≥10 000, major/bidirectional ABO-mismatch, splenomegaly, transfusion reactions, and platelet storage in additive solution negatively affected 2-hour but not 24-hour posttransfusion CCI. The DSA-MFI threshold of 10 000 was corroborated by greater antibody-mediated complement activation and significantly more CCI failures above this threshold, suggesting the usefulness of this value to inform "permissive platelet mismatching" and to optimize CCI. Furthermore, DSA-MFI decreases were deemed feasible by the computer-based algorithm for HLA-platelet selection in a pilot cohort of 8 patients (122 transfusions) evaluated before and after algorithm implementation. When HLA-selected platelets are unavailable, ABO-identical/minor-mismatched platelet concentrates may enhance 2-hour CCI in heavily HLA-alloimmunized patients with platelet transfusion refractoriness.

Graphical abstract

Figure 1.

Patient selection and inclusion/exclusion criteria. (A) Patients with a cPRA of <50% were excluded because our institutional practice is to give random donor platelets in this setting. Generally, alloimmunization is suspected as a cause of PTR after 2 sequential platelet transfusions with poor count increment, however, this is not a strict criterion. Histograms of the total number (B) of platelet transfusions per patient among 2140 total platelet transfusions and the duration (C) of platelet transfusion support (encompassing the time between the first and the last documented transfusion), during the study period. The y-axis in each figure represents patient counts. cPRA, calculated panel reactive antibody; PTR, platelet transfusion refractoriness.

Figure 2.

Patient and component factors affecting 2-hour CCI. Mean 2-hour posttransfusion CCI with 95% confidence interval (y-axis) by patient (A) or component factors (B) analyzed (x-axis); n denotes the number of transfusions affected by the given variable. For time since first HLA-selected component and ratio of HLA-selected products to randomly selected apheresis platelet transfusions (RAP) in previous transfusions, given thresholds were divided by the median values. ∗∗Indicate variables that were statistically significant in univariate and multivariate analyses. ∗Highlights variables that were statistically significant by univariate analysis alone. DSA-MFI impact on CCI is not included in this figure. HSCT, hematopoietic stem cell transplant; PAS, platelet additive solution.

Figure 3.

Impact of cumulative DSA-MFI on 2-hour CCI. (A) Violin plots showing 2-hour posttransfusion CCI (x-axis) by ranges of cumulative DSA-MFI, RAP transfusions, or transfusions with no donor HLA mismatches (y-axis). The horizontal dashed line denotes CCI = 7500. Horizontal bars for each violin plot denote the median for each group, and n for each plot denotes the number of transfusions in each group. ∗Denotes P < .05 in 2-tailed t test compared against 4/4 HLA matched or antigen-negative transfusions. (B) Estimated population mean 2-hour CCI at various values for total MFI conditional on all other predictors included at the linear mixed effects model at their median values; ∗95% confidence band calculated using bootstrap resampling of participants. (C) Association between cumulative DSA-MFI (x-axis) and cumulative C1q MFI compared across 282 transfusions (C1q testing on 58 patient samples). Both axes are log₁₀ scaled. Points are colored by adequate (blue, ≥7500) or inadequate (red, <7500) 2-hour posttransfusion CCI. (D) Mean 2-hour posttransfusion CCI with 95% confidence interval (x-axis) across given ranges of cumulative DSA-MFI by number of mismatched donor HLAs for each DSA-MFI stratum (y-axis).

Figure 4.

The NIH-HLA laboratory platelet component selection algorithm by cumulative DSA-MFI “scores.” (A) The computational framework developed for the allocation of platelet units (Python version 3.10.0). Data manipulation was facilitated using the “Pandas” package. Recipient information, including HLA antibody specificities and their respective immunogenic strengths, was imported for analysis. Additionally, donor information such as unit number, HLA type, unit expiry date, ABO blood group, and Rh type was also imported. The framework conducts an iterative comparison to evaluate compatibility between recipient antibodies and donor HLA types, from which a “compatibility score” or cumulative DSA-MFI is derived. Special attention is paid to certain HLA types (B14, B15, and B40) because of their serological cross-reactivities, with score adjustments for associated antigens. Donors are then ranked according to their compatibility scores, with the system prioritizing those with lower scores to optimize donor-recipient histocompatibility. Violin plots showing cumulative DSA-MFI changes (B) and 2-hour CCI changes (C) before and after algorithm implementation for 61 transfusions each. Horizontal dark lines for each violin plot denote the median for each group, and n for each plot denotes the number of transfusions in each group. Neither of the comparisons was significant at a P < .05 in 2-tailed t tests.