IMPROVED FLOW CYTOMETRIC DETECTION OF HLA ALLOANTIBODIES USING PRONASE

Potential Implications in Renal Transplantation

Abstract

Background. 

Flow cytomeric crossmatch (FCXM) has grown in popularity and has become the “standard of practice” in many programs. Although FCXM is the most sensitive method for detecting alloantibody, the B cell FCXM has been problematic. Difficulties with the B cell FCXMs have been centered around high nonspecific fluorescence background owing to Fc-receptors present on the B cells and autoantibodies. To improve the specificity and sensitivity of the B cell FCXM, we utilized the proteolytic enzyme pronase to remove Fc receptors from lymphocytes before their use in FCXM.

Methods. 

Lymphocytes isolated from peripheral blood, spleen, or lymph nodes were treated with pronase and then used in a three-color FCXM. A total of 167 T- and B cell FCXMs using pronase-treated and untreated cells were performed. Testing used serial dilutions of HLA allosera (22 class I and 6 class II), with the titer of each antibody at one dilution past the titer at which the complement-mediated cytotoxicity anti-human globulin crossmatch became negative.

Results. 

After pronase treatment, the actual channel values of the negative control in both T cell and B cell FCXMs declined from 78±10 to 57±4 (P <0.05) and 107±11 to 49±3 (P <0.00001), respectively. Pronase treatment resulted in improved sensitivity of the T and B cell FCXM in detecting class I antibody by 20% and 80%, respectively. In no instance was a false-positive reaction observed. In this study, pronase treatment improved the specificity of B cell FCXM for detecting class II antibodies from 75% to 100% (P =0.03). In no instance was a false-negative reaction recorded. Lastly, on the basis of these observations we re-evaluated three primary transplant recipients who lost their allografts because of accelerated rejection. One of the patients was transplanted across negative T and B cell FCXM, whereas the other two patients were transplanted across a positive T cell, but negative B cell, FCXM. After pronase treatment, T and B cell FCXMs of each patient became strongly positive, and donor-specific anti-HLA class I antibody was identified in each case.

Conclusion. 

Utilization of pronase-treated lymphocytes improves both the sensitivity and specificity of the FCXM.

INTRODUCTION

The UNOS 1999 Annual Report suggests that, despite significant improvement and progress in immunosuppressive therapy, approximately 8–10% of all transplanted kidneys (i.e., more than1000 kidney per year) are lost within the first 3 months ⁽¹⁾. A significant portion of this graft loss is directly attributable to immunological rejection ⁽¹⁾. Numerous studies, including our own ^{(2, 5)}, have shown that preformed HLA antibodies continue to play a major role in early allograft loss. These antibodies are low titer and generally undetectable even in the most sensitive complement-mediated cytotoxicity (CDC) assays. However, these antibodies are readily detectable by the flow cytometry crossmatch (FCXM) ^(4–6). For this reason, the FCXM has become the “standard of practice” for many transplant centers ^(7–10). Interestingly, since its inception in 1983 by Garovoy et al. ⁽¹¹⁾ and its adaption to a multicolor method by Bray et al. ⁽¹²⁾, the actual FCXM method has remained essentially unchanged. Although interpretation of the T cell FCXM is straightforward and uncomplicated, the B cell FCXM has remained problematic owing to the presence of Fc receptors on the cell surface ⁽¹³⁾. These Fc receptors are also found on the T-cell surface, albeit to a lesser proportion than on the B-cell surface ⁽¹⁴⁾. Normal or irrelevant IgG can bind to such T and B cells, especially when sera contain IgG in an aggregated or complexed form ^(15–16). Consequently, an increase in fluorescence may be interpreted as a positive result when, in fact, it may only indicate the presence of excess irrelevant IgG bound to Fc receptors.

To improve the specificity and sensitivity of FCXM, we have used the proteolytic enzyme pronase to remove Fc receptors from the surfaces of T and B cells before their use in FCXM. In this report, we demonstrate that pronase treatment significantly decreases the nonspecific background fluorescence, thereby enhancing the signal-to-noise separation and consequently increasing the sensitivity and specificity of the FCXM. As demonstrated by clinical outcome, the enhanced dilution of antibody can be significant to renal transplantation.

MATERIAL AND METHODS

Cell preparation.

For cytotoxicity assays, T and B lymphocytes were isolated from peripheral blood, spleen, or lymph nodes using immunomagnetic beads (Dynal Inc., Lake Success, NY) by the method described by Vaidya et al. ⁽¹⁷⁾. For the FCXM assay, mononuclear cells were isolated from either peripheral blood, spleen, or lymph nodes by the standard Ficoll-Hypaque technique and subsequently treated with Lymphokwik (One Lambda, Canoga Park, CA) to remove polymorphonuclear cells and macrophages.

HLA allosera.

A total of 28 distinct human alloantisera were used in this study (22 HLA class I-specific and 6 HLA class II-specific sera). The HLA reactivity of each serum sample was determined by testing against a panel of T cells (n=200) and B cells (n=150) by CDC using Fluorobeads (One Lambda) isolated from peripheral blood lymphocytes. The HLA-specific reactivities of these sera were further confirmed by testing them against HLA-specific microparticles Flow PRA I and II (One Lambda). Additionally, the final crossmatch sera from three primary transplant recipients who lost their allografts because of accelerated rejection, were also studied.

Cytotoxicity assays.

CDC assays included T-cell crossmatches with and without anti-human globulin (AHG) and B-cell crossmatches. In AHG-mediated T-cell crossmatches, T cells and sera were incubated for 30 min at 22°C. The cells were washed three times to remove excess antibodies. After three washes, AHG was added for 1 min before the addition of rabbit complement. The mixture was incubated at 22°C for an additional 60 min before reading and scoring the mixture using a fluorescence microscope. In T-cell crossmatches without AHG, T cells and sera were incubated for 45 min. Rabbit complement was added and the mixture was further incubated for 90 min at 22°C. For B-cell crossmatches, incubation times for cells and sera were 45 min; incubation times for cells with complement were 90 min. Scoring of cytotoxicity of each of the crossmatches was done using criteria defined in the ASHI Standards for the Histocompatibility testing.

Pronase treatment.

Pronase treatment of lymphocytes was done by the method described by Lobo et al. ^{(15, 16)}. In brief, 5–10×10⁶ lymphocytes isolated from peripheral blood, spleen, or lymph nodes were treated with 0.5 mg of pronase (type XIV; Sigma Chemical, St. Louis, MO) (concentration 1 mg/ml, specific activity 4.3×10^-3 units/ml) in a 37°C water bath for 30 min. After incubation, the tubes containing the mixture of lymphocytes and pronase were filled with RPMI without fetal calf serum, and then centrifuged for 1–2 min. at 200×g. The supernatant was discarded, and the cells were adjusted in RPMI to the concentration of 3–5×10⁶ cells/ml. Care was taken to use pronase-treated cells immediately because these cells can regenerate Fc receptors within 8–10 hr ⁽¹⁵⁾. If lymphocytes were used 8–10 hr after their pronase treatment, the storage medium included sodium azide.

Three-color FCXM.

The FCXM was performed using an adaption of the method described by Bray et al. ⁽¹²⁾. One hundred microliters of test serum was added to 3–5×10⁵ lymphocytes suspended in 100 ml of RPMI. The cell serum mixture was incubated at 22°C for 30 min. The cells were then washed three times with RPMI, the supernatant was discarded, and the pellet was resuspended in RPMI. To the pellet, 50 ml (of a 1/50 dilution) of a titered fluoresceinated goat F(ab)₂ anti-human IgG (Jackson Immunoresearch Laboratories, Inc., West Grove, PA) was added. Ten microliters each of anti-CD3 peridinin chlorophyl protein and anti-CD19 phycoerythrin (Becton Dickinson, San Jose, CA) was added to the above mixture to facilitate T- and B-cell discrimination. The mixture was incubated for 20 min in the dark at 4°C. The cells were washed by centrifugation for 5 min at 200×g. The supernatant was discarded, and the cells were resuspended in 200 ml of RPMI and were ready for FCXM analysis. The positive control consisted of high PRA (panel reactive antibodies) serum (sera collected from ≥ five patients with 100% PRAs). The negative controls consisted of (1) pooled heat-inactivated normal human serum and (2) RPMI. Fluorescence was evaluated using a FACScan Flow Cytometer (Becton Dickinson).

The positive cutoff values for each T and B cell FCXM were established as recommended by Cook and Scornik ⁽³³⁾. Briefly, a total of 20 negative control sera known to be devoid of HLA-specific alloreactivities were selected. The majority of the negative control sera (n=17) came from nonsensitized males awaiting renal transplant, whereas the remainder (n=3) were purchased from a commercial source. These sera were tested in T and B FCXMs using lymphocytes from five different donors. The median channel values plus 3 standard deviations for each T and B cell FCXM were considered to be caused by anti-T cell and anti-B cell antibodies, respectively. Without pronase, the positive cutoff values were 30 and 40 median channel shift (MCS) for T and B cell FCXMs, respectively. However, after the pronase treatment, the respective cutoff values were 20 and 30 MCS.

Statistical analysis.

The sensitivity of FCXM was calculated as the probability of a positive FCXM in the presence of known HLA antibodies; the specificity of FCXM was calculated as probability of a negative FCXM in the absence of HLA antibodies. Comparison of sensitivities and specificities of the FCXM using pronase-treated or nontreated cells was done by paired t tests. Chi-square statistics were used to evaluate the influence of pronase on FCXM outcome. For small numbers (n<10), a Fisher exact test was used to evaluate statistical significance. Statistical significance was defined as P ≤0.05.

RESULTS

Selection of allosera.

To test the hypothesis that pronase treatment of lymphocytes improves the overall sensitivity and specificity of the FCXM, 28 HLA-specific allosera were selected for study (22 anti-class I and 6 anti-class II). These sera were reactive to eight different class I and six different class II antigens (Table 1). Analysis with FlowPRA microparticles demonstrated that these sera contained only class I- or class II-reactive antibodies. No serum was used that contained reactivity against both class I and II antibodies.

Table 1: Specificity of HLA allosera

Effect of pronase treatment on known negative control sera.

A total of 20 negative control sera shown to be devoid of HLA-specific alloreactivities were selected. The majority of the negative control sera (n=17) came from nonsensitized males awaiting renal transplantation, whereas the remainder (n=3) were purchased from commercial sources. All sera were shown to be devoid of HLA reactivity by anti-globulin enhanced cytotoxicity as well as FlowPRA microparticles. After pronase treatment, channel values of negative control serum samples for both T and B cell FCXMs declined, on average, from 78±10 to 57±4 (P <0.05) and 107±11 to 49±3 (P <0.00001), respectively (Table 2). Pronase treatment had no statistically significant influence on the fluorescence values of positive control sera (Table 2). However, the decreased background fluorescence after pronase treatment did increase the separation between the negative and positive control samples (i.e., increased signal-to-noise separation).

Table 2: Effect of pronase treatment on control sera

Effect of pronase treatment on sensitivity and specificity of FCXM.

A total of 167 T and B cell FCXMs were performed using serial dilutions of HLA allosera of known specificities. The majority of FCXMs (n=157) were performed using class I antisera, whereas the remainder (n=10) were performed using class II-specific antisera. The titer of each serum used in FCXM was one dilution past the titer at which AHG-CDC crossmatches became negative. The results are shown in Table 3. Of the 157 T and B cell FCXMs, 130 were designed to be true positive (i.e., sera were selected to contain specific antibodies directed at the HLA class I antigens expressed by the cells) and 27 were designed to be true negative (i.e., sera were selected to be devoid of HLA antibodies directed at any of the class I antigens expressed by the cells). There were no instances of false-positive nor false-negative T cell FCXM results (Table 3) (r =1.0 P <0.0001). In contrast, in the absence of pronase treatment, only 80% (104/130) of the FCXMs were positive. The correlation between the presence of class I antibody and the ability of untreated T cell FCXMs to detect antibody was significantly reduced (r =0.6, P <0.0001). However, pronase treatment had no influence in the specificity of T cell FCXM as evidenced by the absence of false-positive reactions. Pronase treatment also exhibited a significant impact on B cell FCXM for its ability to detect class I antibodies (Table 3). When untreated B cells were used in FCXMs, only 20% (26/130) of the B cell FCXMs were positive. However, after the pronase treatment, the sensitivity of a B cell FCXM improved from 20% to 100% (P <0.0001), i.e., no false-negative B cell FCXMs were observed. Pronase treatment did not seem to have an influence on the specificity of B cell FCXM as determined by lack of false-positive reactions. Next, we evaluated the influence of pronase treatment on B cell FCXMs to detect class II antibodies. Of 10 B cell FCXMs performed, 6 were designed to give true positive and 4 were designed to give true negative reactions. There was a perfect correlation score (r =1.0, P =0.01) between the presence of HLA class II antibody and the ability of B cell FCXM to detect the antibody when B cells were treated with pronase. However, when B cells were not treated with pronase, the ability of B cell FCXMs to detect class II antibody declined as measured by the decline in r values from 1.0 to 0.8. The specificity of B cell FCXM declined by 25% (Table 3). No false-negative reactions were observed either with or without the pronase treatment. As expected, each of 10 T cell FCXMs performed with class II antisera were negative. This fact further supports the finding that the class II antisera did not contain class I reactivities.

Table 3: Sensitivity and specificity analysis of FCXMs

The false-negative T and B cell FCXMs (Table 3) using class I antisera were further evaluated. MCS values of each of the false-negative T cell (n=26) and the B cell (n=104) FCXMs were plotted in a pair-wise fashion (Fig. 1, left and right). The box shows the central location and the dispersion of MCS values of T cell (Fig. 1, left) and B cell (Fig. 1, right) FCXMs for class I antibodies. The diamond adjacent to the box shows the average values of MCS and 95th percentile confidence interval around the average. The average MCS of false-negative T cell FCXMs was 14±8. However, after the pronase treatment, these T cell FCXMs became unequivocally positive as evidenced by MCS values of 43±15. This difference was statistically significant (P <0.0001). Similarly, the average MCS value of false-negative B cell FCXMs was 17±14 using untreated cells. In contrast, after pronase treatment, there was a statistically significant increase in the average MCS value of B cell FCXMs (64±18, P <0.0001). These results demonstrated that pronase treatment significantly increased signal-to-noise separation, thereby increasing the sensitivity of the FCXMs.

Figure 1: MCS values of each of the T cell (n=26) and B cell (n=104) FCXMs with or without pronase, were plotted in a pair-wise fashion. The box shows the central location and dispersion of MCS of T and B cell FCXM for class I antibodies. Average MCS with versus without pronase treatment for the T cell FCXMs are 43±15 vs. 14±8 (P<0.0001) and for the B cell FCXMs, 64±18 vs. 17±14 (P<0.0001). The diamond adjacent to the box shows the 95th percentile confidence interval around the average. For the T cell FCXM with pronase treatment, the 95% confidence interval is from 68 to 18 MCS; without pronase treatment the 95% confidence interval is 28 to 0 MCS. For the B cell FCXM, the 95% conficence interval with pronase treatment is 95 to 35 MCS, and without the pronase the confidence interval is 40 to −5 MCS.

Figure 2 shows representative examples of T and B cell FCXMs using serial dilutions of various HLA antisera. In Figure 2A, anti-HLA-A3 antibody was tested against HLA-A3-positive T (Fig. 2A, a) and B (Fig. 2A, b) lymphocytes. T cell FCXMs were positive at all three dilutions regardless of the pronase treatment, however, the strength of positivity as measured by MCS was significantly greater with pronase-treated T cells compared with the untreated T cells. Pronase treatment had a major impact on the outcome of B cell FCXM (Fig. 2A, b). Without pronase treatment, B cell FCXMs were only weakly positive at a 1:2 dilution, becoming negative at two subsequent dilutions. However, after pronase treatment, the B cell FCXM became strongly positive and remained positive for the two subsequent dilutions. Figure 2B, a and b, shows similar but more pronounced impact of pronase treatment on T and B cell FCXM using an anti-HLA-B60 serum tested against B60-positive lymphocytes. Before pronase treatment, the T cell FCXM was borderline positive and became negative at subsequent dilutions. In contrast, after pronase treatment, the T cell FCXM became strongly positive and remained positive at 1:4 and 1:8 dilutions. In contrast, the B cell FCXM (Fig. 2B, b) was completely negative before pronase treatment, but became unequivocally positive after pronase treatment and remained positive for two more dilutions. Figure 2C, a and b, demonstrates the influence of pronase treatment on T and B cell FCXM using an anti-DQ2 antibody. As expected, T cell FCXMs (Fig. 2C, a) were negative, confirming the fact that anti-DQ2 antisera had no class I reactivities, although the B cell FCXMs were positive regardless of the pronase treatment (2C, b), and the MCS values of pronase-treated B cells were consistently greater at each dilution.

Figure 2: FCXMs were set up using various HLA antibodies; anti-HLA-A3 (Fig. 2A), anti-HLA-B60 (Fig. 2B) and anti-HLA-DQ2 (Fig. 2C). In T cell FCXM, MCS≥30 is considered positive. In B cell FCXM MCS≥40 is considered positive.

Impact of pronase treatment on donor-specific FCXMs.

On the basis of the above observations, three primary transplant patients who rejected their renal allografts because of early accelerated rejections within 3–7 days of their transplantations were re-evaluated. These patients were transplanted across negative CDC crossmatches. One patient (patient 1, Table 4) had negative T and B cell donor-specific FCXMs, whereas two of the patients had positive T cell FCXM but negative B cell FCXM (patients 2 and 3, Table 4). Because HLA antigens are present at higher density on B cells than on T cells, antibodies causing positive T cell FCXM, but negative B cell FCXM, were presumed to be non-HLA and therefore considered irrelevant ⁽¹⁸⁾. Retrospective FCXMs were setup for all three patients using their original kidney donors’ lymphocytes treated with or without pronase. Table 4 shows the results of retrospective FCXMs with and without pronase. The repeat of the final, prospective crossmatch showed results similar to the pretransplantation crossmatch. However, the T and B cell FCXMs for each of the patients became positive when pronase-treated lymphocytes were used. Furthermore, donor-specific anti-HLA class I antibody was confirmed in each case when these final crossmatch sera were tested using FlowPRA microparticles (data not shown).

Table 4: Impact of pronase treatment on donor-specific FCXM

DISCUSSION

The use of the FCXM in clinical transplantation has grown steadily since its first introduction almost two decades ago ^(19–24). The FCXM is considered by many to be the most sensitive method for identifying HLA antibodies. Specifically, the FCXM can detect antibodies that are not identifiable by other cytotoxic or ELISA-based techniques ⁽²⁵⁾. However, there is a great deal of debate regarding the clinical significance of antibodies detected only by FCXM in renal transplantation. Several transplant centers have found a strong association between a positive FCXM and early allograft loss in both primary as well as regraft recipients ^{(7–11, 19–24)}. In contrast, there are transplant programs that consider the FCXM to be “too sensitive” because antibodies detected by flow cytometry have not been associated with decreased renal allograft survival ^{(26, 27)}. One reason for such a controversy is that the FCXM assay, as currently performed, lacks standardization. The lack of standardization of the FCXM assay became quite evident when Scornik et al. ⁽²⁸⁾ published the summary of ASHI-CAP FCXM proficiency survey results in 1997. Per their report, the FCXM methodology varied considerably from one laboratory to another. Consequently, there was little agreement in terms of interpretation of a FCXM as positive or negative for a known IgG, HLA class I antibody. Although the level of consensus was higher when HLA class I antibody concentration was greater than 70%, the consensus level fell significantly when the antibody concentration was lower. The lack of agreement in the FCXM outcome was also reported by a multicenter study conducted in the United Kingdom ⁽²⁹⁾. The lack of consensus in the interpretation of FCXM outcome could be for several reasons. First, there was not a consensus as to “cut-off values” for assigning a positive FCXM. For positive T cell FCXMs, cut-off values can vary from MCS≥10 up to MCS≥50 using the same channel scale ⁽⁷⁾. The interpretation of positive B cell FCXMs is even more problematic because the cut-off values for positive B cell FCXMs vary from MCS≥40 up to MCS≥150 ^(7–10).

A second reason for discrepancies in the FCXM is that individual laboratories utilize different methods. Data published by Scornik et al. ⁽²⁸⁾ showed that for laboratories that use a dilution of patient sera, their overall sensitivity is decreased compared with laboratories that use “neat” or undiluted serum in the crossmatches. Lastly, an additional reason for the variability in the FCXM is the presence of Fc receptors on a small number of T cell and all B cell subsets ^(13–14). These Fc receptors can bind irrelevant (non-HLA specific) IgG present in the negative control sera or the test patient sera. The binding of irrelevant IgG in the negative control sera would most likely increase background fluorescence, thereby decreasing the signal-to-noise separation and potentially producing a false-negative FCXM. In contrast, binding of irrelevant IgG in the patient’s serum may produce a false-positive FCXM.

To address these issues, we evaluated the role of pronase treatment as a means to eliminate Fc receptors from the cell surface of T and B lymphocytes before their use in the FCXM assay. Pronase is a type IV proteolytic enzyme that is produced by Streptomyces griseus K-1. It has at least three distinct caseinolytic activities and one aminopeptidase activity. Pronase removes Fc receptors or molecules bearing Fc receptor-like domains from the cell surface. Fc receptors are present on some T cells but on all B cells ^{(13, 14)}, which is the most likely reason why pronase treatment has more pronounced effects on the outcome of B cell rather than T cell FCXMs. It is interesting to note that although pronase does not cleave CD19, it does cleave the CD20 receptor from cell surfaces ^{(15, 16)}, because CD20 has an Fc molecule-like structure. Therefore, care should be taken not to use the incorrect CD markers when using pronase. HLA class I and class II molecules do not possess Fc receptor-like domains, rendering them resistant to pronase activity. Our data showed that after pronase treatment, not a single FCXM was false-negative in the presence of HLA-specific antibodies (Table 3). These results are consistent with the fact that pronase does not affect cell-surface markers such as CD3, CD19, and HLA class I and class II molecules.

Before using pronase, we evaluated different technical approaches to reduce the background in B cell FCXM with little success. In one approach we used different quantities of sera in FCXM. We performed dose-response studies using fixed number of lymphocytes (n=5×10⁵) and fixed amount of secondary antibodies, while varying the amount of sera used from 50 μl up to 150 μl in 25-μl increments (data not shown). We found that the background fluorescence did not vary with the volume of serum but it did vary with donor lymphocytes. Some donor lymphocytes were more prone to background than others. We also evaluated the influence of lymphokwik in inducing high background as suggested in one study ⁽³²⁾. We saw no difference in the background fluorescence whether or not lymphokwik was used (data not shown). Similarly, there is some concern that pooled normal human sera used as a negative control may cause high background when compared with sera obtained from a single source. We compared our background fluorescence generated by pooled normal human sera with two single-source normal human serum samples (generous gifts from Dr. Daniel Cook of Cleveland Clinic and Dr. Robert Bray of Emory University, respectively) in 20 parallel FCXM. Our results showed no difference in background fluorescence (data not shown) whether we used pooled normal human serum as a negative control or serum sample obtained from a single source. These observations lead us to the hypothesis that the source high background particularly in the B cell FCXM is probably not due to quantity of sera, the source of the sera, nor monocyte contamination. The problem may be due to the Fc receptors present on the B cells.

Our data support the hypothesis that pronase-treated lymphocytes improved the sensitivity and specificity of FCXM. This finding was evident for both T and B cells, as well as for class I and class II antibodies. As hypothesized, the improvement in sensitivity was largely due to a significant decrease in background fluorescence of the negative control sera. As a consequence, there was a marked reduction in the false-negative FCXMs, especially in B cell FCXMs. Several investigators have found the B cell FCXM to be unreliable because of the inability to detect HLA class I antibodies even though HLA class I antigens have been reported to be at higher density on B cells than on T cells ⁽¹⁸⁾. Antibodies causing positive T cell but negative B cell FCXMs are presumed to be non-HLA antibodies and therefore clinically irrelevant for transplant consideration. Pronase treatment eliminates such ambiguous FCXM results. Our data also demonstrated that after pronase treatment, there was a marked improvement in the specificity of the FCXM, particularly in detecting class II antibodies.

Our results show that the presence of graft-damaging alloantibodies was either not detected or equivocal when we used lymphocytes without pronase treatment (Table 4). However, in retrospective FCXMs using pronase-treated donor cells, we demonstrated unequivocally the presence of donor-specific anti-HLA class I antibodies. In all three instances, these antibodies were most likely the cause of the early accelerated allograft destruction ^{(2–5, 20, 30, 31)}.

In summary, pronase treatment is a simple and rapid assay that enhances signal-to-noise separation, thereby improving both the sensitivity and specificity of the T and B cell FCXM assay.