Long-term in vitro reactivity for human leukocyte antigen antibodies and comparison of detection using serum versus plasma

Abstract

Background: Human leukocyte antigen (HLA) antibodies are a possible cause of transfusion-related acute lung injury (TRALI), and fluorescent bead assays are often used for antibody detection. Serum is the manufacturer's recommended sample, but plasma may be easier to obtain for studies of HLA antibody prevalence and TRALI case investigations.

Study design and methods: Specimens were obtained from 44 multiparous females positive for the presence of HLA antibodies by lymphocytotoxicity testing at least 13 years prior and from 1000 contemporary blood donors. Screening tests were performed using a multiplex bead-based assay. In addition to comparing results obtained with paired plasma and serum samples, the effects of storage at 4 degrees C for 1 week and of multiple freeze-thaw cycles were evaluated.

Results: Of 42 evaluable subjects with HLA antibodies documented more than 13 years earlier, only 1 showed loss of detectable antibodies, with 39 (93%) positive in the screening assay for HLA Class I and 24 (57%) positive in the screening assay for HLA Class II antibodies. In 968 evaluable contemporary donors, 291 screened positive for the presence of HLA Class I and 206 for HLA Class II antibodies using a low assay cutoff. Screening test concordance using paired plasma and serum samples was high, particularly for subjects with higher-level antibodies. Refrigeration of samples for 1 week did not significantly affect assay results, while repeated freeze-thaw cycles caused a decrement in signal level.

Conclusion: Serum and plasma samples gave concordant results in the majority of cases, particularly for specimens with higher-level antibodies. High-level HLA antibodies were present in most individuals for more than 13 years.

Fig. 1. Concordance between plasma and serum screen test results

Screening tests (LSM12) were performed in singlicate on 1,000 paired plasma and serum samples from the contemporary cohort of blood donors. NBG ratios obtained in serum are plotted vs. the plasma NBG ratio in the left hand column for HLA class I (A), HLA class II (B), and MICA (C) antibodies. The formula for a linear regression curve and R² are listed in each plot. The right hand column presents Bland-Altman plots of the data, with the difference in the serum and plasma NBG ratios plotted against the mean of the two NBG ratios. The thick line indicates the mean bias and the two dashed lines depict the mean bias +/− 1.96 times the standard deviation of the NBG difference. The two samples with the highest class II plasma NBG ratios (B) were also positive using serum samples and were excluded from the figure and analysis as the outliers distorted the line of best fit.

Fig. 2. Screen test results for plasma or serum samples that were positive in only one type of sample

The NBG ratio is shown for all samples that tested positive by screening in only one of the two sample types for HLA class I (A) (plasma n=54, serum n=70) or class II (B) (plasma n=30, serum n=47) antibodies. An assay cut-off of NBG ≥ 1.7 was used for plasma samples and NBG ≥ 2.2 was used for serum samples. Most samples that yielded a positive result in only one sample type were clustered near the assay cut-off, though higher signals were seen in a few samples.

Fig. 3. Effects of refrigerated storage and freeze-thaw cycles on assay signal

Screening tests (LSM12) are shown for triplicate samples of plasma and serum aliquots from 15 subjects that had been frozen and thawed once and assayed immediately or stored at 4 °C for one week. Results are also shown for four additional samples that were frozen and thawed for two additional cycles and tested in duplicate or triplicate. Results are displayed as percentage and absolute decrement in NBG ratio, and error bars represent the standard error. Panel A shows HLA class I and panel B shows class II results.