Answers are grounded in this paper and cite their sources.
After the demonstration that the presence of donor-specific antibodies (DSA) in pretransplantation serum is associated with hyperacute rejection of the kidney allograft (1), it has become mandatory to examine recipient’s serum for DSA before each kidney transplantation. Testing for DSA has progressed from complement-dependent cytotoxicity assays using donor’s cells as targets to a battery of solid-phase assays. In solid-phase testing, human leukocyte antigens (HLA) are fixed to a solid substrate, incubated with patient’s serum, and subsequently tested for bound anti-HLA antibodies. Whereas, the first-generation solid-phase assays were designed to detect pan-IgG anti-HLA antibodies, recent refinements permit detection of C1q-fixing or C4d-depositing anti-HLA antibodies (2). These assays may better resolve the relevance of the DSA in the clinical setting. The results from Luminex platform-based solid-phase assay are reported as mean fluorescence intensity (MFI). Variations in MFI due to technical factors such as day-to-day variations or technologist-based variations need consideration in the interpretation of the results of the assay (3, 4). Another potential contributor to the MFI variability is lot-to-lot variations. Our earlier report in an abstract form in year 2010 (5) and the elegant study of Gandhi et al. (6) reported in 2013 have focused on the lot-to-lot variability. The results from our investigation documenting in detail the lot-dependent variability in MFI are reported in this letter. We used a pan-HLA-ABC–specific monoclonal antibody (mAb) W6/32 (eBioscience, San Diego, CA) that binds human β2-microglobulin to investigate variability in MFIs among three consecutive lots (LS1A04 lots 5–7) of HLA Class I single antigen beads (SAB) (One Lambda, Canoga Park, CA). The binding of W6/32 mAbs to the SAB was quantified using trimmed MFI values. The trimmed MFI is the mean MFI calculated by the data analyzer software (Luminex 100 IS; Luminex, Austin, TX) after removal of the lower and upper 5% of the fluorescent intensities of the bound fluorochrome. The MFI was extracted directly from the instrument’s output file with no further modification of the MFI values. Pearson correlation analysis was used to compare SAB MFI measured on two different occasions using the same lot (intralot comparison) or to compare the SAB MFI measured using two different lots (interlot comparison) and Spearman’s rank–order correlation was used for intralot and interlot comparison of the ranks of the SAB MFI. The Bland–Altman plots were used to calculate the difference in SAB MFI within the same lot or between two different lots (7). Figure 1 illustrates the near-perfect correlation (Pearson correlation coefficient squared [R2]=0.99; 95% confidence intervals, 0.99–0.99) observed after comparison of lot 5 SAB MFI with lot 5 SAB MFI (Fig. 1A) and lot 6 SAB MFI with lot 6 SAB MFI (Fig. 1C). Figure 1B shows the less-than-perfect correlation (R2=0.41; 95% confidence interval, 0.50–0.74) observed between lot 5 SAB MFI and lot 6 SAB MFI. Spearman rank–order correlation (rs) analysis of SAB MFI ranks yielded the following correlations: rs=0.99 for intralot comparison of SAB MFI ranks of lot 5, rs=0.99 for intralot comparison of lot 6, and rs=0.58 for interlot comparison of SAB MFI ranks of lot 5 with SAB MFI ranks of lot 6.FIGURE 1: Lot-to-lot variations in MFI. A pan-HLA-ABC–specific mAb W6/32 that binds human β2-microglobulin was used to investigate variability in MFIs among different lots (LS1A04 lots 5–7) of HLA class I SABs. The binding of W6/32 mAbs to the SAB was quantified using trimmed MFI values. The trimmed MFI is the mean MFI calculated by the data analyzer software after removal of the lower and upper 5% of the fluorescent intensities of the bound fluorochrome. The MFI was extracted directly from the instrument’s output file with no further modification of the MFI values. Statistical analysis of the MFI of the SAB and SAB MFI rank order showed a near-perfect correlation between SAB MFI within a lot (A, lot 5 vs. lot 5, Pearson R 2=0.99; C, lot 6 vs. lot 6, R 2=0.99) and only a modest relationship among the SAB MFI from two different lots (B, lot 5 vs. lot 6, R 2=0.41). The intralot comparisons (A and C) were performed by the same technologist using the same instrument and measurements were made 6 days apart. The interlot comparison (B) was performed by the same technologist using the same instrument and on the same day. All Pearson R 2 yielded P values less than 0.0001. The Bland–Altman plots show clearly that the interlot variability (difference in MFI; E) exceeds that of intralot variability in MFI values (D and F).The Bland–Altman plots show that the interlot variability in the MFI (difference in MFI) of SAB (Fig. 1E, lot 5 vs. lot 6) exceeds the variability observed by intralot comparison (Fig. 1D, lot 5 vs. lot 5; Fig. 1F, lot 6 vs. lot 6). To minimize the impact of technologist- or instrument-based variations on interlot or intralot comparisons, the same technologist used the same instrument to generate the data illustrated in Figure 1. To minimize the day-to-day variations, the interlot (Fig. 1B, lot 5 vs. lot 6) comparison was performed on the same day and intralot comparisons were carried out 6 days apart. The interlot variability observed by comparing lot 5 SAB MFI or SAB MFI rank with lot 6 SAB MFI and SAB MFI rank was also observed when lot 5 was compared with lot 7 (R2=0.35; rs=0.50) and when lot 6 was compared with lot 7 (R2=0.52; rs=0.69). Again, the intralot comparison of lot 7 yielded near perfect correlations (R2=0.99; rs=0.98). It is not known whether W6/32 mAbs bind all HLA class I molecules effectively and whether the lot-to-lot variability identified with the use of W6/32 mAbs is applicable to clinical samples. We therefore investigated whether lot-to-lot variability is also found with serum from patients awaiting kidney transplantation. Our testing of sera from highly sensitized patients with SAB from lots 5 and 6 showed that several SABs are assigned to different categories when MFI thresholds of less than 1000, 1001 to 2999, and 3000 or more are used as cut points to assign the SAB to a particular category (Table 1).TABLE 1: Lot-to-lot variations observed with sera from patients awaiting kidney transplantationaTransplant programs currently list unacceptable antigens for the United Network for Organ Sharing waitlist and for paired kidney exchange registries based on MFI values. MFI values in sequential sera are also used to gauge the success of desensitization protocols. In the posttransplantation setting, recipients are screened for anti-HLA-IgG antibodies to identify the fluctuations in anti-HLA-IgG antibodies. In all these clinical settings, interpretation of MFI ascertained using different lots might pose challenges. The variability identified in our study, however, is based on the commercial kit from one vendor. It would be important to examine the variability, if any, in the products from other vendors. What might be a practical way to deal with lot-dependent variations? The inclusion of a control (reference) serum with each lot is a potential approach for normalizing MFI values, but reference sera are not usually provided with the commercial kits. The histocompatibility laboratories could prepare a mixture of well-characterized sera and use the results from the sera to normalize MFI values across the lots. Alternatively, W6/32 mAbs could be used. W6/32 mAbs as reference serum for normalizing MFI HLA class I antigens has certain merits because of ready availability, consistency inherent to mAbs, and opportunities for validation across laboratories. A similar strategy for HLA class II antigens however needs to be developed. Rex Friedlander 1 Prabhakar Putheti1,2 Elder Diaz1 Arvind Menon1 Blanca Ponce1 Thangamani Muthukumar2 Vijay K. Sharma1,2 Darshana Dadhania1,2 Manikkam Suthanthiran1,2 1 Immunogenetics and Transplantation Center The Rogosin Institute New York, NY 2 Department of Medicine Weill-Cornell Medical Center New York, NY
No takes yet. Share an insight, caveat, or question.
Friedlander et al. (2013) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: