Capillary C4d Deposition in Kidney Allografts: A Specific Marker of Alloantibody-Dependent Graft Injury

Patients

Between June 1998 and September 1999, 226 cadaveric kidney transplants were performed in our unit. For 124 patients (54.9%), at least one allograft biopsy was performed during the follow-up period (up to July 2001). For a cohort of 58 of the 124 recipients who underwent biopsies, cryopreserved donor splenocytes and serum samples obtained before transplantation and at the time of biopsy were available for retrospective serologic testing. These patients were enrolled in this study. One of the included recipients received a combined kidney/pancreas transplant, and another received a combined kidney/heart transplant. The median patient age in our study cohort (male/female ratio, 32/26) was 51 yr [interquartile range (IQR), 37 to 60 yr]. Twelve recipients had received a retransplant. The median HLA mismatch was 3 (IQR, 2 to 3), and the median cold ischemia time was 13 h (IQR, 9 to 18).

Immunosuppression

Forty of the 58 study patients received initial baseline immunosuppressive therapy consisting of cyclosporin A (CsA) (dose initially adjusted to yield trough levels between 200 and 400 ng/ml), mycophenolate mofetil (2 g/d, administered orally), and steroids. Nine recipients received rapamycin (initial dose, 2 mg/d) together with CsA and steroids. Two retransplanted and sensitized patients [cytotoxic panel-reactive antibody (PRA) reactivity, >10%], one combined kidney/heart transplant recipient, and three recipients of suboptimal grafts (non-heart-beating donor or donor age of >60 yr) received induction therapy with rabbit or horse antilymphocyte serum (ALS). Another three recipients underwent an induction protocol with the anti-interleukin-2 receptor antibody Daclizumab (Hoffman-La Roche, Basel, Switzerland) (n = 1) or Basiliximab (Novartis AG, Basel, Switzerland) (n = 2). Histologically proven acute rejection episodes and some cases of Banff borderline lesions were initially treated with steroid bolus therapy (dexamethasone administered at 100 mg/d for 3 d). Steroid-resistant graft dysfunction was treated with antilymphocyte polyclonal antibodies (ALS) or, for one patient, antilymphocyte monoclonal antibody (mAb) (OKT3; Jansen, Raritan, NJ). Thirteen patients received tacrolimus rescue therapy. As previously described (15), two patients with characteristic features of humoral rejection (capillary C4d staining, intracapillary granulocytes, and severe graft dysfunction) received immunoadsorption (IA) therapy in addition to ALS (both patients also exhibited features of cellular rejection). Two recipients with clinical and histopathologic diagnoses of hemolytic uremic syndrome (positive C4d staining for one patient) underwent a change to tacrolimus treatment (CsA-induced hemolytic uremic syndrome was suspected) and plasma-exchange therapy.

Renal Allograft Pathologic Analyses

Biopsies were performed because of persistent oliguria or anuria or acute functional impairment, after exclusion of post- and prerenal causes of graft dysfunction and toxic CsA (>400 ng/ml) or FK506 (>20 ng/ml) levels. During the follow-up period (up to July 2001), 113 renal allograft biopsies were performed for the 58 included patients. Thirty-three of these patients underwent more than one biopsy. Biopsies were performed a median of 19 d after transplantation (range, 3 to 681 d). Nineteen biopsies were performed >3 mo after transplantation.

Histopathologic evaluation and C4d staining were performed with formalin-fixed, paraffin-embedded sections. Specimens were stained with hematoxylin and eosin, periodic acid-Schiff stain, methenamine-silver, and the trichrome stain acidic fuchsin-Orange G. Lesions in allograft biopsies were classified according to the definitions provided by the Banff 97 working classification of renal allograft pathologic features (16).

For detection of C4d, we used a novel polyclonal anti-C4d antibody (C4dpAb; Biomedica, Vienna, Austria) generated by immunization of a rabbit with a 15-mer peptide corresponding to amino acids 1242 to 1256 of C4. The generation and characterization of C4dpAb are described elsewhere in detail (14). Specific binding of C4dpAb to purified C4 and tryptic C4d was demonstrated by Western blot analysis. Furthermore, antibody specificity was proven in blocking experiments with tissue sections. Preincubation of C4dpAb with purified C4 or the peptide used for immunization completely abolished reactivity with frozen sections of a representative C4d⁺ kidney allograft. In contrast, irrelevant control peptides or other complement factors (C3 or C5) did not affect the staining results. Identical peritubular staining patterns for C4d with C4dpAb and an anti-C4d mAb (Quidel, Alkmaar, Netherlands) were demonstrated in a series of renal biopsies (12 allografts and 25 normal native kidneys) for which both paraffin-embedded and frozen sections were available. Immunostaining of paraffin-embedded sections with C4dpAb revealed the same peritubular staining pattern as that observed for corresponding frozen sections stained with anti-C4d mAb (14). For immunohistochemical detection of C4d on formalin-fixed, paraffin-embedded sections, we used indirect immunoperoxidase staining. Sections (2 μm) were deparaffinized, and endogenous peroxidase activity was blocked with hydrogen peroxide/methanol. Antigen retrieval was performed by pressure-cooking (at 1 bar) for 10 min in citrate buffer (pH 6.0). Endogenous biotin was blocked by using a biotin blocking kit (Vector Laboratories, Burlingame, CA). After a 30-min incubation with C4dpAb (5 μg/ml), bound IgG was observed by using the Supersensitive Kit (BioGenex, San Ramon, CA), according to the protocol provided by the manufacturer. The intensity of endothelial C4d staining was classified as weak, moderate, or strong. Some of the biopsies with low-intensity C4d staining demonstrated an inhomogeneous distribution of C4d [only a few C4d⁺ peritubular capillaries (PTC)]. In biopsies with moderate or strong C4d staining, virtually all capillaries stained positive.

Immunologic Methods

Pretransplant cytotoxic crossmatch testing was performed according to the protocol of the Eurotransplant Organization, using the standard microcytotoxicity technique described by Terasaki and McClelland (17). For inactivation of IgM, sera were pretreated with dithiothreitol. All study patients exhibited negative cytotoxic crossmatch findings before transplantation. Pretransplant cytotoxic PRA reactivity was assessed by using a panel of cells from 50 phenotyped donors.

Fifty-eight pretransplant and 113 posttransplant serum samples obtained at the time of renal biopsy were retrospectively tested for the presence of alloantibodies. For FCXM testing, cryopreserved spleen cells isolated from donor spleens by density gradient centrifugation were used. We used a three-color technique. Donor cells (4 × 10⁵) were incubated with undiluted serum at room temperature for 1 h. The cells were washed and then incubated with a mixture of pretitered, FITC-conjugated, goat anti-human IgG (Accurate Chemical and Scientific Co., Westbury, NY), phycoerythrin-labeled CD3 mAb, and peridin-chlorophyll-A-protein-conjugated CD19 mAb (Becton Dickinson, San Jose, CA) at 4°C for 30 min. Fluorescence intensity was measured by using a FACSCalibur flow cytometer (Becton Dickinson). The lymphocyte population was gated according to forward- and side-scatter characteristics. Recipient reactions against T cells (T cell FCXM) or B cells (B cell FCXM) were analyzed by measuring FITC binding to the cells. According to previous studies (18,19), the FCXM was considered positive when the mean fluorescence intensity was greater than the mean + 2 SD of values for normal control samples (serum samples from four different nonsensitized healthy volunteers). Positivity was graded as FCXM_high if mean fluorescence intensity levels were twofold higher than mean control fluorescence intensity levels.

The FlowPRA screening test, a recently established, HLA-specific, flow cytometric technique (20), was performed according to the guidelines provided by the supplier (One Lambda, Inc.). Using fluorescent microbeads coated with purified HLA antigens (either class I or II) from 30 different cell lines, this assay allows measurement of anti-HLA class I and class II PRA reactivity. In brief, 20 μl of serum were incubated with 5 μl of FlowPRA class I (FL1-30, lot 6) and 5 μl of FlowPRA class II (FL2-30, lot 7) beads. After a 30-min incubation period at room temperature, the beads were washed twice and then stained for 30 min with appropriately pretitered, FITC-conjugated, anti-human IgG (Fcγ; One Lambda, Inc.). Fluorescence intensity was measured by flow cytometry. The major population of beads was gated in the forward- versus side-scatter dot plot. FlowPRA class I and II beads were discriminated in the FL2 histogram. The marker was set according to the negative control sample, i.e., serum provided by the company (FL-NC; One Lambda, Inc.). The percentage of HLA class I- or class II-coated beads shifted to the right of the cutoff point represents the percentage of FlowPRA class I or class II reactivity. Sera with ≥5% FlowPRA class I and/or II reactivity were considered anti-HLA antibody-positive. Anti-HLA class I antibody-positive sera contained 26% (median; IQR, 10 to 83%) FlowPRA class I reactivity, and anti-HLA class II antibody-positive sera contained 37% (median; IQR, 18 to 50%) FlowPRA class II reactivity.

Sera with ≥5% FlowPRA reactivity were tested for specific HLA class I (FlowPRA specific HLA class I antibody detection test, FL1SP, lot 006; One Lambda, Inc.) and/or HLA class II (FlowPRA specific HLA class II antibody detection test, FL2SP, lot 004; One Lambda, Inc.) antibodies. The respective bead panels (each composed of four groups containing eight different colored beads with different FL2 fluorescence properties) were composed of 32 microbeads coated with purified HLA (class I or II) antigens. Beads were incubated with serum and stained with FITC-conjugated anti-human IgG as described for the screening test. The major population of beads was gated in the forward- versus side-scatter dot plot. Different beads were discriminated in the FL2 histogram. Markers were set according to the negative control sample (negative control serum, FL-NC; One Lambda, Inc.). Positivity was scored according to the percentage of beads shifted (0 to 20%, negative; 21 to 100%, positive). The specificity of HLA antibodies was calculated from staining patterns, using the software provided by the manufacturer (One Lambda, Inc.).

Statistical Analyses

Continuous data are presented as the median and IQR (range from the 25th to the 75th percentile). Percentages were calculated for dichotomous variables. Sensitivity, specificity, positive predictive value, and negative predictive value were calculated by using standard formulae. Comparisons between groups were performed by using the χ² test, the Mann-Whitney U test, or the Kruskal-Wallis test, as appropriate. A multivariate logistic regression model was used to assess the independent effects of baseline variables on renal outcomes (serum creatinine levels at 12 mo of >1.755 versus <1.755 mg/dl; patients undergoing dialysis, creatinine levels of >1.755 mg/dl; patient deaths excluded). Variables demonstrating a trend toward a difference in graft function (P < 0.2) were entered into the model as possible predictor variables. Results are presented as the odds ratio and 95% confidence interval. The Cox-Snell pseudo-r² value was calculated for assessment of the predictive value of the model, and the Hosmer-Lemeshow test was used to assess the model fit. All P values are two-sided. P values of <0.05 are reported as statistically significant. A commercially available computer program (SPSS for Windows, version 10.0; SPSS Inc., Chicago, IL) was used for all statistical calculations.

Association of Capillary C4d Staining with Posttransplant Serologic Results

Twenty-four of the 113 renal allograft biopsies evaluated (21.2%) demonstrated C4d deposits in PTC (C4d_PTC⁺). The prevalence of positive FCXM findings among the 113 tested serum samples obtained at the time of biopsy was 60.2% (i.e., 68 FCXM⁺ sera; T cell reactivity only, n = 8; B cell reactivity only, n = 15; T plus B cell reactivity, n = 45). In a comparison of capillary C4d staining with FCXM results, 21 of the 24 C4d_PTC⁺ biopsies (87.5%) and 47 of the 89 C4d_PTC⁻ biopsies (52.8%) were associated with FCXM positivity. The 113 posttransplant sera were further tested for the presence of anti-HLA alloantibodies by using the FlowPRA screening test. Fifty-five sera (48.7%) demonstrated ≥5% FlowPRA reactivity (anti-HLA class I only, n = 7; anti-HLA class II only, n = 19; anti-HLA class I and II, n = 29). Forty-one sera demonstrated both FCXM positivity and ≥5% FlowPRA reactivity. Fourteen sera exhibited ≥5% FlowPRA reactivity but negative FCXM results, and 27 sera demonstrated FCXM positivity in the absence of detectable anti-HLA antibodies. Twenty-one of the 24 C4d_PTC⁺ biopsies (87.5%) were associated with ≥5% FlowPRA reactivity (20 of these sera were also FCXM⁺). The specificity, sensitivity, positive predictive value, and negative predictive value of C4d staining and FlowPRA analysis (FCXM testing was defined as the standard method) are presented in Table 1. Using the FlowPRA specific HLA antibody detection test, we were able to identify HLA specificities for one or more mismatched donor HLA class I and/or class II antigens in 19 of the 55 posttransplant sera with ≥5% FlowPRA reactivity. Eleven of those sera were associated with C4d positivity.

Effects of Capillary C4d Staining and Posttransplant Serologic Results on Kidney Allograft Outcomes

Thirty-five of the 58 study patients (60.3%) experienced one or more biopsy-proven rejection episodes, according to the Banff classification. Thirteen patients demonstrated Banff type I, 18 Banff type II, and four Banff type III rejection (for patients for whom more than one biopsy was obtained, the rejection episode of highest Banff grade is reported). Histologic diagnoses for the remaining 23 patients were borderline lesions (n = 10), acute tubular necrosis (ATN) (n = 7), chronic allograft nephropathy (n = 2), thrombotic microangiopathy (n = 1), donor-derived injury (n = 2), and minor/nonspecific changes (n = 1). For 28 recipients, graft dysfunction (27 cases of classified Banff rejection and one Banff borderline lesion) was resistant to high-dose steroid treatment, and antilymphocyte antibody therapy was administered. Clinical outcomes for patients with steroid-resistant graft dysfunction are presented in Table 2. Five recipients lost their grafts because of refractory rejection. For another two patients, severe C4d_PTC⁺ acute allograft rejection was reversible with combined IA and antithymocyte globulin (ATG) therapy (15). Clinical and immunologic data for patients with severe graft dysfunction are summarized in Table 3. During the follow-up period, five deaths occurred (three resulting from severe septicemia, one resulting from cerebral hemorrhage, and one after aortic valve replacement).

Patients were first divided into two groups according to C4d staining results, i.e., patients with positive C4d staining in at least one allograft biopsy (C4d_PTC⁺, n = 16) and C4d_PTC⁻ recipients (n = 42). As calculated in univariate analyses, serum creatinine levels at 12 mo were significantly higher for C4d_PTC⁺ patients (2.83 mg/dl; IQR, 1.93 to 4.2 mg/dl) than for C4d_PTC⁻ patients (1.6 mg/dl; IQR, 1.32 to 2.15 mg/dl; P = 0.001); for calculations, a serum creatinine concentration of 5 mg/dl was presumed for patients undergoing dialysis, and patient deaths (i.e., two C4d_PTC⁺ and three C4d_PTC⁻ recipients) were excluded from analysis. Furthermore, C4d_PTC⁺ patients experienced more episodes of refractory rejection than did C4d_PTC⁻ patients (four of 16 versus one of 42 patients, P = 0.02). The incidences of Banff rejection were similar for the two groups (10 of 16 versus 25 of 42 patients). No significant difference was observed between the two groups with respect to the incidence of steroid-resistant graft dysfunction (nine of 16 versus 19 of 42 patients). Two C4d_PTC⁺ patients exhibited steroid-resistant rejection associated with characteristic histopathologic features of acute humoral rejection (granulocytes in PTC), which was successfully reversed with IA plus ALS therapy. For multivariate logistic regression analysis, serum creatinine levels at 12 mo were taken as the dependent variable (serum creatinine levels of <1.755 versus >1.755 mg/dl; patients undergoing dialysis were included in the calculation). After adjustment of the model for baseline variables with P values of <0.2 (Table 4), i.e., donor age, cold ischemia time, and delayed graft function, C4d staining was observed to be independently associated with inferior allograft function (P = 0.02) (Table 5).

In a subsequent analysis, patients were subdivided according to both C4d staining and FCXM results. We defined three patient subsets, i.e., the group of 16 recipients with (presumably alloantibody-mediated) capillary C4d staining (C4d_PTC⁺; 13 patients also exhibited FCXM positivity) and two C4d_PTC⁻ patient groups, which were discriminated on the basis of posttransplant FCXM results, i.e., C4d_PTC⁻/FCXM⁺ recipients (n = 22) and C4d_PTC⁻/FCXM⁻ recipients (n = 20). Immunologic results obtained for these patient groups are presented in Table 6. The proportion of patients classified as FCXM_high (definition given in Materials and Methods) was higher for the C4d_PTC⁺ group than for the C4d_PTC⁻/FCXM⁺ group (12 of 16 versus 10 of 22 patients). Univariate analyses revealed significant differences among the three patient groups with respect to serum creatinine levels at 12 mo (Figure 1). In addition, the single C4d_PTC⁻ case of refractory allograft rejection was classified as C4d_PTC⁻/FCXM⁺. In the multivariate logistic regression analysis (confounding factors were donor age, cold ischemia time, and delayed graft function), significantly higher serum creatinine levels at 12 mo were observed for C4d_PTC⁺ patients, compared with C4d_PTC⁻/FCXM⁻ patients (Table 7). Differences in allograft function between C4d_PTC⁻/FCXM⁺ and C4d_PTC⁻/FCXM⁻ patients, however, did not achieve statistical significance.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1. Box plot indicating the median serum creatinine levels at 12 mo (interquartile range and range). For patients undergoing dialysis, a serum creatinine level of 5 mg/dl was presumed. For univariate analysis, the Kruskal-Wallis test was used. The five patient deaths [peritubular capillary C4d deposit-positive (C4d_PTC⁺), n = 2; C4d_PTC⁻/flow cytometric crossmatch-positive (FCXM⁺), n = 2; C4d_PTC⁻/FCXM⁻, n = 1] were excluded from analysis.

Association of Pretransplant Serologic Findings with Capillary C4d Staining

Eleven of the 16 C4d_PTC⁺ recipients (68.8%) exhibited positive pretransplant FCXM findings (T and/or B cell FCXM). Positive pretransplant FCXM findings were significantly less common among C4d_PTC⁻ patients (7 of 42 patients, 16.7%; P < 0.0001). Furthermore, recipient presensitization detected by PRA testing was more common among C4d_PTC⁺ patients than among C4d_PTC⁻ patients (≥5% FlowPRA reactivity, 10 of 16 versus 15 of 42 patients, P = 0.08; ≥5% cytotoxic PRA reactivity, six of 16 versus five of 42 patients, P = 0.026).