Peritransplant Immunoadsorption for Positive Crossmatch Deceased Donor Kidney Transplantation


Corresponding author: Georg A. Böhmig,


Various desensitization protocols were shown to enable successful living donor kidney transplantation across a positive complement-dependent cytotoxicity crossmatch (CDCXM). Positive crossmatch transplantation, however, is less well established for deceased donor transplantation. We report a cohort of 68 deceased donor renal allograft recipients who, on the basis of broad sensitization (lymphocytotoxic panel reactivity ≥40%), were subjected to a protocol of peritransplant immunoadsorption (IA). Treatment consisted of a single session of immediate pretransplant IA (protein A) followed by posttransplant IA and antilymphocyte antibody therapy. Twenty-one patients had a positive CDCXM, which could be rendered negative by pretransplant apheresis. Solid phase HLA antibody detection revealed preformed donor-specific antibodies (DSA) in all 21 CDCXM-positive and in 30 CDCXM-negative recipients. At 5 years, overall graft survival, death-censored graft survival and patient survival were 63%, 76% and 87%, respectively, without any differences between CDCXM-positive, CDCXM-negative/DSA-positive and CDCXM-negative/DSA-negative recipients. Furthermore, groups did not differ regarding rates of antibody-mediated rejection (24% vs. 30% vs. 24%, p = 0.84), cellular rejection (14% vs. 23% vs. 18%, p = 0.7) or allograft function (median 5-year serum creatinine: 1.3 vs. 1.8 vs. 1.7 mg/dL, p = 0.62). Our results suggest that peritransplant IA is an effective strategy for rapid desensitization in deceased donor transplantation.


Recipient sensitization represents a leading cause of kidney allograft rejection and loss (1,2). Moreover, creating a barrier to transplant access, preformed humoral alloreactivity often leads to unacceptably long waiting times without a suitable crossmatch-negative organ offer (1,2). In recent years, various desensitization strategies comprising plasmapheresis, intravenous immunoglobulin (IVIG) and/or anti-CD20 antibody rituximab, have been demonstrated to enable successful living donor kidney transplantation across the barrier of a positive crossmatch (3–11). Moreover, for broadly sensitized patients awaiting a deceased donor kidney transplant, repeated administration of high-dose IVIG, with or without rituximab, was shown to improve transplant rates by decreasing levels of HLA sensitization (5,9,12). Such treatment, however, is not always effective and may bring risks associated with prolonged immunosuppression on dialysis, without the certainty of a suitable transplant offer. Currently, there is only limited experience with crossmatch conversion immediately before deceased donor transplantation (13–15). In this context, immunoadsorption (IA) may represent an attractive strategy for pronounced and selective alloantibody removal within the short time window between an organ offer and transplant surgery. In two earlier studies, IA with staphylococcal protein A columns has been examined as a strategy for rapid desensitization (13,14). Higgins et al. (13) reported successful crossmatch conversion and prevention of hyperacute rejection by one or two extended pretransplant IA sessions. However, in this study, considerable graft loss rates (five of nine patients) were observed among recipients transplanted across a positive complement-dependent cytotoxicity crossmatch (CDCXM) (13). We have modified this initial protocol by excluding patients from transplantation in whom a positive CDCXM could not be rendered negative by IA treatment of 6 L patient plasma (14). The rationale behind this threshold was to minimize immunological risks and to avoid an unacceptable prolongation of cold ischemia times as a result of extended treatment. In addition, posttransplant IA treatment was added to prevent alloantibody rebound. In a first report of 40 sensitized recipients, who had been subjected to peritransplant IA, we could demonstrate favorable intermediate-term transplant outcomes in nine CDCXM-positive recipients, without a difference to sensitized CDCXM-negative patients treated with the same protocol (14).

Extending our initial experience, we here provide a thorough analysis of 68 broadly sensitized deceased donor kidney allograft recipients subjected to peritransplant IA. Based on CDC panel-reactive antibody (PRA) levels ≥40%, these recipients were subjected to preemptive IA without extended desensitization on the waiting list. Our analysis included 21 recipients with a positive baseline CDCXM that could be rendered negative by immediate pretransplant IA. Applying bead array technology for HLA alloantibody detection, we evaluated the incidence and clinical impact of pre- and posttransplant donor-specific HLA alloreactivity in the particular context of IA-based desensitization.

Materials and Methods


Between January 1999 and December 2008, 68 broadly sensitized recipients of a deceased donor kidney transplant (current CDC-PRA reactivity ≥40%) were subjected to peritransplant IA. Median CDC-PRA was 73%[interquartile range (IQR): 56–86%]. Forty-three recipients have received a second, thirteen a third, seven a fourth and two a fifth allograft. All transplants were performed within the Eurotransplant region, whereby 29 kidneys had been allocated via the Acceptable Mismatch Program (16). Baseline characteristics are summarized in Table 1.

Table 1.  Baseline characteristics and immunosuppressive treatment
VariablesCDCXM+/DSA+ (N = 21)CDCXM-/DSA+ (N = 30)CDCXM-/DSA- (N = 17)p-Value
  1. CDCXM, complement-dependent cytotoxicity crossmatch; CDC-PRA, complement-dependent cytotoxicity panel-reactive antibody; IA, immunoadsorption; IQR, interquartile range; MMF, mycophenolate mofetil; N, number.

Recipient age (years), median (IQR)48 (36–57)43 (36–51)47 (39–52)0.55
Female sex, N (%)9 (43)10 (33)7 (41)0.76
Prior kidney transplantation, N (%)20 (95)28 (93)17 (100)0.56
%CDC-PRA, median (IQR)82 (62–94)73 (51–92)68 (51–78)0.10
HLA mismatch (A, B and DR), median (IQR)3 (2–4)3 (2–3)3 (1–4)0.49
Waiting time to transplant (years), median (IQR)2.4 (1.3–4.3)2.8 (1.8–3.9)3.0 (2.2–4.7)0.68
Donor age (years), median (IQR)32 (17–52)46 (36–54)43 (37–52)0.11
Cold ischemia time (h), median (IQR)18 (14–22)17 (12–22)17 (15–22)0.90
Peritransplant IA
 Immediate pretransplant IA session
   Processed plasma volume (L), median (IQR)9 (7–11)8 (7–9)8 (8–11)0.57
   Duration of treatment (h), median (IQR)5 (4–7)5 (4–6)4 (4–7)0.87
 Posttransplant IA treatment
   Number of IA sessions, median (IQR)8 (5–11)10 (7–13)10 (6–13)0.21
   Duration of treatment course (days), median (IQR)13 (6–24)22 (14–27)17 (12–33)0.09
Prophylactic depleting anti-lymphocyte antibody, N (%)19 (90)27 (90)17 (100)0.41
IL-2 receptor antibody induction, N (%)2 (10)3 (10)0 (0)0.41
Maintenance immunosuppression
 Tacrolimus, N (%)12 (57)14 (47)9 (53)0.76
 Cyclosporin, N (%)9 (43)16 (53)8 (47)0.76
 MMF, N (%)19 (91)27 (90)17 (100)0.41
 Azathioprine, N (%)2 (10)3 (10)0 (0)0.41

Peritransplant IA

The protocol of peritransplant IA has earlier been described in detail (14). Immediately before planned transplantation, transplant candidates were subjected to IA treatment with staphylococcal protein A (Fresenius HemoCare, St. Wendel, Germany). Before initiation of treatment and after desorption of 6 L of plasma, serum samples were obtained for sequential CDCXM testing, and IA was continued until two to three plasma volumes were processed. Patients with a negative baseline CDCXM or a positive CDCXM that could be rendered negative by pretransplant IA proceeded to transplantation. Following transplantation, all patients continued to receive IA every 1–3 days until stabilization of kidney function or graft loss, respectively (maximum duration of treatment: 7 weeks).

Twenty-one recipients had a positive baseline CDCXM that could be rendered negative by pretransplant IA. Forty-seven recipients were CDCXM-negative already before treatment. Fourteen prior transplant offers for 10 of the patients could not be realized because of a persistent positive CDCXM. Baseline characteristics were well balanced among CDCXM-positive, CDCXM-negative/DSA-positive and CDCXM-negative/DSA-negative recipients, with the exception of a trend toward a lower donor age and a higher CDC-PRA reactivity in CDCXM-positive patients (Table 1). Patient groups did not differ with respect to intensity and duration of immediate pretransplant IA (Table 1). Accordingly, there was no increase in cold ischemia time among CDCXM-positive patients. Moreover, groups did not differ regarding the number of posttransplant IA sessions, even though there was a trend toward a shorter duration of IA treatment in CDCXM-positive recipients (Table 1).

Immunosuppression and treatment of rejection

As shown in Table 1, 63 recipients were given preemptive therapy with a depleting antilymphocyte antibody [Thymoglobuline® (IMTIX-Sangstat, Lyon, France), Lymphoglobuline® (IMTIX-Sangstat) or ATG Fresenius (Fresenius Medical Care Ltd., Bad Homburg, Germany)]. The remaining five recipients received IL-2 receptor antibody induction with Basiliximab (Novartis Pharma, Basel, Switzerland) or Daclizumab (Hoffmann La Roche AG, Basel, Switzerland). Antibody therapy was administered shortly after IA treatment to minimize adsorption to protein A columns. Maintenance immunosuppression consisted of a calcineurin inhibitor (cyclosporine A or tacrolimus), mycophenolate mofetil or azathioprine and steroids (Table 1). Ganciclovir or valganciclovir were administered as prophylaxis against cytomegalovirus infection, trimethoprim plus sulfamethoxazole as prophylaxis against pneumocystis pneumonia.

Episodes of acute C4d-positive antibody-mediated rejection (AMR, N = 15) occurred within the first weeks after transplantation and were commonly diagnosed under preemptive IA treatment. Only one recipient had AMR (type I; biopsy at day 50) after IA treatment was terminated. This case was complicated by postrenal allograft failure, and after percutaneous nephrostomy, kidney function gradually improved without the necessity of intensified immunosuppression. In the other patients, treatment consisted in a prolongation of the posttransplant IA treatment course. Accordingly, among patients with acute AMR a longer duration of the IA course [median 27 days (IQR: 16–38 days) vs. median 16 days (IQR: 10–23 days; p = 0.003)] and a higher number of IA sessions [median number: 12 (IQR: 9–14 days) vs. 9 (IQR: 6–12); p = 0.016] were reported. In recipients with pure AMR (N = 13), graft dysfunction could be reversed by continued posttransplant IA, with the exception of three recipients with C4d-positive thrombotic microangiopathy. One of these recipients was subjected to plasma exchange and tacrolimus rescue, however without success. Two recipients with concomitant cell-mediated rejection (Banff type II) were, in addition to IA, subjected to high-dose steroids and ATG. One of the two recipients did not respond to treatment.

Antibody detection

CDCXM was performed applying the standard microcytotoxicity technique originally described by Terasaki and McClelland (17). Tests were carried out in the presence or absence of dithiotreitol to distinguish between IgG and IgM reactivities. For CDC-PRA, sera were tested on a panel of mononuclear cells obtained from 30 to 50 phenotyped donors.

Sera obtained immediately before pretransplant IA and 3–9 months after transplantation were subjected to retrospective serological testing on a Luminex platform. Anti-HLA single antigen (SA) reactivity was detected using LABScreen® Single Antigen assays (HLA class I: LABScreen® Single Antigen HLA Class I Antibody Detection Test—Combi; HLA class II: LABScreen® Single Antigen HLA Class II Antibody Detection Test—Group 1; One Lambda, Canoga Park, CA, USA) according to the manufacturer's protocol. Results of SA reactivities were recorded as mean fluorescence intensity (MFI). Only beads with MFI levels above 500 were considered to be positive. In addition, to eliminate false positive test results due to nonspecific antibody binding, for each SA bead, test thresholds were chosen according to negative control bead binding and the results obtained with five nonbinding negative control sera as earlier described (18). Test results were analyzed in the context of donor and recipient HLA class I (A, B, Cw) and class II (DR and DQ) typing results. Moreover, virtual classes I and II PRA in an Austrian population was calculated using a software tool at the Eurotransplant website ( (19).


According to local standard, allograft biopsies were performed for renal dysfunction after exclusion of toxic calcineurin inhibitor levels or post- and prerenal causes of graft dysfunction. Protocol biopsies were not routinely performed. Forty-seven recipients were subjected to at least one indication biopsy during follow-up, 38 within the first 3 months. Overall, 103 biopsies (one to five per patient) were performed after a median of 36 days (range 1–1968). Rejection was diagnosed and classified according to the Banff classification (20,21). All biopsy specimens were stained for C4d on paraffin sections applying a polyclonal rabbit anti-C4d reagent (Biomedica, Vienna, Austria) (22). Fifteen patients were diagnosed for C4d-positive acute AMR (type I: N = 6, AMR type II: N = 9) after a median of 14 days (range: 7–50 days), five recipients had C4d-positive chronic AMR (uncovered by indication biopsy 147–1568 days posttransplantation), two of them with a prior episode of reversible acute AMR. Thirteen patients had acute cellular rejection (type I: N = 3, type II: N = 10) diagnosed after a median of 22 (4–147) days.


Chi-square or Fisher's exact tests were used to compare proportions. For comparison of continuous data, nonparametric methods were applied (Mann–Whitney test, Kruskal–Wallis test). Paired nonparametric testing was performed using the Wilcoxon test. Kaplan–Meier analysis was used to calculate graft survival and the Mantel Cox log-rank test was applied to compare survival between groups. Multivariate Cox proportional hazards analysis was performed to adjust for baseline imbalances. A two-sided p-value <0.05 was considered as statistically significant. Statistical calculations were performed using SPSS for Windows, version 12.0 (SPSS Inc., Chicago, IL, USA).


Sixty-eight broadly sensitized recipients of a deceased donor kidney transplant were subjected to peritransplant IA without additional desensitization on the waiting list. In this cohort, 5-year death-censored graft survival, overall graft survival and patient survival were 76%, 63% and 87%, respectively. Causes of graft loss were acute (N = 4) or chronic C4d-positive AMR (N = 4), postoperative vascular thrombosis without any features of cellular or humoral rejection (N = 5), glomerulonephritis (N = 2) or nonbiopsied chronic dysfunction (N = 3). Causes of death were disseminated cancer (N = 2; one patient died of gastric cancer and one of melanoma), death from unknown cause (N = 2), postoperative hemorrhagic shock after bleeding of the perigraft region (N = 1) or intracerebral hemorrhage (N = 1). In addition, two patients died in an accident.

Transplant Outcome in Relation to Baseline Serology

Detailed results of baseline CDC and solid phase HLA antibody testing are provided in Table 2. Twenty-one of the 68 sensitized patients had a positive baseline CDCXM that could be rendered negative by pretransplant IA. All 21 CDCXM-positive recipients had preformed DSA (HLA class I plus class II: N = 11; HLA class I: N = 5; HLA class II: N = 5). Forty-seven recipients had a negative baseline CDCXM. Thirty of these recipients (64%) were identified to have preformed DSA (HLA class I plus II DSA: N = 13; HLA class I: N = 9; HLA class II: N = 8). With the exception of two CDCXM-negative patients (one with, one without DSA), all included recipients had preformed reactivity to an array of nondonor antigens (not shown). There was no difference between CDCXM-positive/DSA-positive and CDCXM-negative/DSA-positive patients regarding HLA class specificity of DSA or the number of targeted donor HLA antigens. However, among CDCXM-positive recipients, significantly higher MFI levels were recorded for HLA class II DSA (Table 2).

Table 2.  Results of pretransplant solid phase HLA antibody testing–donor-specific alloreactivity
Pretransplant DSACDCXM+/DSA+ (N = 21)CDCXM-/DSA+ (N = 30)p-Value
  1. CDCXM, complement-dependent cytotoxicity crossmatch; DSA, donor-specific antibody; IQR, interquartile range; MFI, mean fluorescence intensity; N, number.

  2. 1The number of targeted antigens was recorded for samples with detectable donor-specific HLA class I and/or class II alloreactivity.

  3. 2For each DSA-positive patient, the peak MFI level detected for targeted donor HLA class I and/or class II antigen-coated beads was recorded and included in the analysis.

HLA class I and/or II, patient number (%)21 (100)30 (100)
 Targeted antigens1, N, median (IQR)2 (1–4)2 (1–3)0.46
 MFI2, median (IQR)3868 (2393–5566)2196 (1570–3780)0.013
HLA class I, patient number (%)16 (76)22 (73)0.82
 Targeted antigens1, N, median (IQR)1 (1–2)1 (1–2)0.88
 MFI2, median (IQR)1888 (1272–4742)2772 (1570–3780)0.76
HLA class II, patient number (%)16 (76)21 (70)0.63
 Targeted antigens1, N, median (IQR)1 (1–2)1 (1–2)0.82
 MFI2, median (IQR)3978 (2462–5419)1782 (1210–2961)0.003

In 10 study patients, 14 prior transplant offers could not be realized because of a persistent positive CDCXM. Pre-IA serum samples did not differ from baseline samples obtained in the 21 cases of successful CDCXM conversion regarding the number of identified HLA class I and/or class II DSA [median number: 3 (IQR: 2–3) vs. 2 (1–4); p = 0.76] or maximum MFI levels of detected DSA [median MFI: 3097 (IQR: 1373–4507) vs. 3868 (2393–5566); p = 0.2].

For analysis of clinical outcomes, patients were divided into three groups according to baseline serology (CDCXM-positive and DSA-positive; CDCXM-negative and DSA-positive; CDCXM-negative and DSA-negative). Groups did not differ regarding death-censored graft survival (p = 0.91; Figure 1), overall graft survival (p = 0.49) or patient survival (p = 0.49). Moreover, Cox regression analysis adjusting for donor age and CDC-PRA (trend toward a difference between groups) did not reveal any significant effect of baseline serology on transplant survival [CDCXM-positive patients: adjusted HR for death-censored survival: 0.75, 95% confidence interval: 0.27–2.56; p = 0.75; CDCXM-negative/DSA-positive patients: 0.91, 95% confidence interval: 0.24–3.53; p = 0.91 (reference group: DSA-negative patients)].

Figure 1.

Comparative analysis of death-censored graft survival in CDCXM-positive/DSA-positive (N = 21), CDCXM-negative/DSA-positive (N = 30) and CDCXM-negative/DSA-negative (N = 17) kidney allograft recipients subjected to peritransplant IA. The Mantel Cox log-rank test was used to compare Kaplan–Meier graft survival between groups.

As shown in Table 3, groups did not differ regarding rates of C4d-positive acute or chronic AMR, respectively. There was also no difference regarding AMR type or concomitant cellular rejection. Moreover, patient groups showed comparable rates of acute cellular rejection, and a comparison of groups regarding active Banff lesion scores revealed no significant differences (Table 3). As shown in Table 4, rates of delayed graft function exceeded 35% without any difference between groups. Moreover, groups had comparable serum creatinine values or levels of urinary protein excretion 1, 3 or 5 years after transplantation, and there was no difference regarding the occurrence of viral or bacterial infectious complications (Table 4).

Table 3.  Transplant rejection and histological lesions in indication biopsies, according to baseline serology
ParametersCDCXM+/DSA+ (N = 21)CDCXM-/DSA+ (N = 30)CDCXM-/DSA- (N = 17)p-Value
  1. AMR, antibody-mediated rejection; CDCXM, complement-dependent cytotoxicity crossmatch; DSA, donor-specific antibody; IQR, interquartile range; N, number.

  2. 1All episodes of acute AMR were diagnosed within the first 3 months.

  3. 2Two of the recipients with chronic AMR had an episode of early acute AMR, which responded to treatment.

  4. 3For patients with more than one biopsy, the maximum lesion score was recorded.

Acute cellular rejection, N (%)*3 (14)7 (24)3 (18)0.7
Banff type I, N030 
Banff type II, N343 
C4d-positive acute and/or chronic AMR, N (%)5 (24)9 (30)4 (24)0.84
 Acute AMR, N (%)14 (19)7 (23)4 (24)0.92
   Banff type I, N132 
   Banff type II, N342 
   Additional cellular rejection, N111 
 Chronic AMR, N (%)21 (5)3 (10)1 (6)0.75
Banff active lesion scores
 Patients scored, N3132212 
   Glomerulitis (g), median (IQR)0 (0–0)0 (0–2)0 (0–0)0.36
   Interstitial inflammation (i), median (IQR)1 (0–2)1 (0–2)0 (0–1)0.22
   Tubulitis (t), median (IQR)0 (0–2)2 (0–2)0 (0–2)0.67
   Intimal arteritis (v), median (IQR)0 (0–1)0 (0–0)0 (0–0)0.34
Table 4.  Clinical outcomes in relation to baseline serology
VariablesCDCXM+/DSA+ (N = 21)CDCXM-/DSA+ (N = 30)CDCXM-/DSA- (N = 17)p-Value
  1. CDCXM, complement-dependent cytotoxicity crossmatch; CMV, cytomegalovirus; DSA, donor-specific antibody; IQR, interquartile range; N, number; SrCr, serum creatinine.

  2. 1Delayed graft function was defined as the need for dialysis within the first posttransplant week.

  3. 2Patients on dialysis were assumed as having a serum creatinine of 5 mg/dL, and included in nonparametric statistical analysis.

  4. 3Patients on dialysis were excluded from analysis of urinary protein excretion.

  5. 4One patient developed CMV pneumonia.

  6. 5Polyoma virus nephropathy occurred in one patient and was proven by a biopsy performed for chronic dysfunction at 3 years. The transplant maintained stable function with an SrCr of 2.2 mg/dL after 8 years.

Delayed graft function1, N (%)8 (38)14 (47)8 (47)0.74
SrCr (mg/dL)2, median (IQR)
 1 year1.5 (1.2–4.5)1.6 (1.3–2.3)1.5 (1.3–1.8)0.86
 3 years1.3 (1.1–5.0)1.6 (1.3–4.3)1.5 (1.1–5.0)0.78
 5 years1.3 (1.1–5.0)1.8 (1.4–5.0)1.7 (1.2–5.0)0.62
Protein excretion (g/L)3, median (IQR)
 1 year<0.05 (<0.05–0.17)<0.05 (<0.05–0.12)<0.05 (<0.05–0.07)0.44
 3 years<0.05 (<0.05–0.22)<0.05 (<0.05–0.38)<0.05 (<0.05-<0.05)0.18
 5 years<0.05 (<0.05–0.28)0.06 (<0.05–0.27)0.27 (0.07–0.57)0.48
Urinary tract infection, N (%)4 (19)11 (37)4 (24)0.35
Bacterial pneumonia, N (%)2 (10)4 (13)2 (12)0.92
CMV infection, N (%)40 (0)4 (13)1 (6)0.19
Herpes zoster, N (%)0 (0)1 (3)0 (0)0.53
Polyoma virus nephopathy, N (%)50 (0)1 (3)0 (0)0.53

Finally, we were interested whether the 18 patients with acute and/or chronic C4d-positive AMR differed from patients without AMR (N = 50) with respect to pretransplant alloreactivity patterns. With the exception of a marginally higher median number of targeted HLA class II donor antigens in patients with AMR (p = 0.05), we found no differences regarding baseline CDCXM, incidence and specificity of DSA, number of targeted donor HLA antigens or MFI values detected for DSA (Table 5).

Table 5.  Pretransplant serology in relation to the occurrence of biopsy-proven acute and/or chronic C4d-positive AMR
VariablesAcute and/or chronic AMR (N = 18)No AMR (N = 50)p-Value
  1. AMR, antibody-mediated rejection; CDCXM, complement-dependent cytotoxicity crossmatch; DSA, donor-specific antibody; IQR, interquartile range; MFI, mean fluorescence intensity; N, number

  2. 1The number of targeted antigens was recorded for samples with detectable donor-specific HLA class I and/or class II alloreactivity.

  3. 2For each DSA-positive patient, the peak MFI level detected for targeted donor HLA class I and/or class II antigen-coated beads was recorded and included in the analysis.

CDCXM positive, N (%)5 (32)16 (28)0.74
Pretransplant DSA
 HLA class I and/or II, patient number (%)14 (78)37 (74)0.75
   Targeted antigens1, N, median (IQR)2 (1–3)2 (1–3)0.86
   MFI2, median (IQR)2771 (1860–4056)2431 (1733–4663)0.92
 HLA class I, patient number (%)11 (61)27 (54)0.60
   Targeted antigens1, N, median (IQR)1 (1–2)1 (1–2)0,53
   MFI2, median (IQR)2875 (1487–3985)2077 (1527–3773)0.54
 HLA class II, patient number (%)8 (44)29 (58)0.32
   Targeted antigens1, N, median (IQR)2 (1–2)1 (1–2)0.05
   MFI2, median (IQR)2494 (1571–3482)2355 (1472–4707)0.85

Incidence and Clinical Impact of Posttransplant HLA Alloreactivity

For 52 patients, posttransplant serum samples collected 3–9 months after transplantation were available for HLA antibody testing. Thirty-nine of these 52 recipients had preformed DSA whereas 50 patients had antibodies to other HLA antigens. After transplantation, 23 patients had alloreactivity against donor HLA antigens, 48 patients against other HLA antigens. Out of the 39 patients with pretransplant DSA, 20 became DSA-negative, whereas 4 of the 13 DSA-negative patients turned DSA-positive. We observed a significant lower number of targeted donor antigens and lower MFI values when comparing DSA reactivities between the pre- and posttransplant sera (Figure 2A). In addition, while there was apparently no considerable change in the number of recipients with nondonor alloreactivity, we observed a modest, but statistically significant decrease in levels of the virtual PRA (Figure 2B).

Figure 2.

Analysis of donor-specific HLA alloreactivity and virtual PRA before and after transplantation. (A) The number of targeted donor antigens and MFI (DSA) (for each serum the highest MFI value was recorded) is shown for pre- versus posttransplant samples containing detectable DSA. (B) Pre- and posttransplant virtual PRA was calculated according to individual reactivity patterns identified by HLA SA testing. Box plots indicate median, IQR and range. Mild outliers are indicated as open dots, extreme outliers as asterisks. For statistical comparisons, nonparametric testing was applied.

Detection of persistent or de novo DSA after transplantation was not associated with adverse death-censored graft survival as compared to patients without posttransplant DSA (p = 0.61, data not shown). There was also no difference regarding rates of acute and/or chronic C4d-positive AMR (DSA-positive vs. DSA-negative patients: 26% vs. 24%, p = 0.87) or cellular rejection (17% vs. 21%, p = 0.76). Among the four patients with de novo DSA, one developed acute C4d-positive graft dysfunction (graft loss after 18 months due to nonbiopsied chronic dysfunction). The remaining three patients, however, had no rejection episode, and maintained stable graft function.


This study suggests that peritransplant IA is a safe and efficient desensitization strategy for deceased donor kidney transplant recipients at extensive immunological risk. A major finding was that IA-based rapid desensitization enabled successful transplantation across a positive CDCXM. In addition, preformed DSA uncovered by bead array technology, which were also detected in the majority of CDCXM-negative subjects, did not have any impact on rejection or graft loss rates.

Although our observation of favorable allograft outcomes in DSA-positive recipients supports high efficiency of our approach, it is important to mention that this study has several limitations. First, without a head-to-head comparison to a nontreated control group of subjects at equal immunological risk, our observational approach is not suitable to provide definitive proof of efficiency. Nevertheless, the extensive rejection risk in CDCXM-positive recipients (23) may preclude the design of such a randomized controlled trial. Second, despite a 10-year recruitment period, only 68 patients were included, a result of the limited number of broadly sensitized patients on the waiting list. These relatively small sample sizes, which, still exceed those reported in other evaluations of recipient desensitization, suggest a careful interpretation of our data. Finally, according to study design, also CDCXM-negative (and DSA-negative) patients were included, and it remains unclear whether these recipients, who may be at comparably low immunological risk, really benefit from IA treatment.

A novel aspect of our study is the successful use of rapid desensitization immediately before transplant surgery, without additional immunosuppression on the waiting list. In this respect, our strategy substantially differs from other protocols, which rely on extended immunomodulatory treatment in advance of transplantation (3–12). One could argue that the high efficiency of our protocol may have resulted from the use of IA technology. A major advantage over plasmapheresis may be that IA enables treatment of extended plasma volumes without the need for fresh frozen plasma to substitute for albumin or coagulation factors (24). In support of high antihumoral efficiency, IA has earlier been shown to be effective in AMR treatment, without the need for adjunctive treatment measures, such as IVIG or rituximab (25,26). Interpreting our results, it has to be noted that our regimen was designed to allow transplantation only in patients in whom a positive CDCXM could be rendered negative upon desorption of 6 L of plasma. This threshold was chosen to avoid extensive prolongations of treatment resulting in unacceptable cold ischemia times. Moreover, it was our intention to minimize immunological risks by excluding transplants with excessive levels of complement-activating donor reactivity, which would have necessitated treatment of very large plasma volumes. In this respect, an earlier study has to be mentioned where high-volume IA for CDCXM conversion (up to 43 L plasma treatment) was associated with markedly prolonged cold ischemia times (13). Another explanation for efficiency of our protocol could be the inclusion of posttransplant IA treatments, which, in conjunction with modulation of B cell immunity by antilymphocyte antibody treatment (27,28), may have counteracted the deleterious effects of early alloantibody rebound.

Solid phase DSA were frequently detected among CDCXM-negative subjects, and there was a considerable overlap between CDCXM-positive and CDCXM-negative samples regarding antibody binding intensity and number of targeted antigens. These data suggest that demarcation between CDCXM-positive and CDCXM-negative samples may not necessarily mean high versus low DSA. Indeed, distinct qualitative properties, such as the complement fixing capability of alloantibodies, may not solely be a function of antibody binding strength (29,30). Notably, absolute MFI levels of DSA in our recipients were apparently lower than those earlier reported for crossmatch-positive samples (31). Such differences could be due to intercenter variabilities in assay sensitivity, which may preclude a direct comparison of absolute MFI levels. An interesting observation was that in cases of a persistent positive CDCXM absolute MFI levels did not differ from those recorded for cases of successful crossmatch conversion, suggesting that MFI values detected for DSA may not be useful for prediction of the success of our protocol.

Following transplantation, a considerable proportion of our patients lost antidonor HLA reactivity, whereas nondonor alloreactivity was apparently less affected. One possible explanation for this interesting observation could be adsorption of DSA to the allograft. Indeed, in support of a role of DSA adsorption, some authors have demonstrated elution of DSA from rejecting allografts (32,33), and several reports have shown a marked increase in circulating HLA alloreactivity following transplant nephrectomy (34,35). Alternatively, downregulation of donor-specific reactivity in desensitized transplant recipients was also suggested to be related to active suppression of donor-specific B cells (36).

Interestingly, some patients had persistent donor-specific anti-HLA reactivity, and it was a remarkable finding that detection of posttransplant DSA did not predict higher rejection or graft loss rates. Similar observations have been made by others, both in HLA- and ABO-incompatible kidney transplantation, where stable graft function was described despite circulating donor-specific reactivity or complement split product deposition in surveillance biopsies, respectively (7,37–40). Even though speculative, our finding of persistent alloreactivity could reflect transplant accommodation, a state of alloantibody-induced resistance of the allograft to immunological damage (41,42). However, a major limitation of our study is the lack of serial protocol biopsies, so that no definitive conclusions can be drawn regarding the actual relevance of DSA in our cohort. One could also argue that DSA detection indicates occult AMR leading to late graft injury. Indeed, in recent reports, surveillance biopsies following recipient desensitization by plasmapheresis/IVIG have uncovered substantial rates of early sublinical AMR (38,43,44). Of note, analysis of our indication biopsies revealed no difference between groups regarding active lesions, such as glomerulitis, a lesion discussed to be related to ongoing antibody-mediated rejection processes. A low number of late allograft biopsies, however, precluded a valid statistical comparison with respect to distinct chronic lesions related to humoral injury, such as allograft glomerulopathy.

A considerable proportion of our patients experienced AMR (27%), which definitely exceeded AMR rates reported for our overall transplant population (45). A similar increase in rejection rates has also been demonstrated for other desensitization strategies, such as plasmapheresis plus IVIG, where rejection rates between 30% and 50% have been reported (11). Moreover, we are aware that the use of C4d as a major rejection criterion may be too insensitive to detect all episodes of AMR. Indeed, emerging evidence suggests a potential role of C4d-negative AMR (46). Accordingly, the actual incidence of humoral rejection in our cohort may have been underestimated. It can be speculated that the observed increased AMR rate culminates in chronic allograft injury. However, small sample sizes and the lack of protocol biopsies may preclude definitive conclusions regarding the actual impact of IA on antibody-mediated injury in the long term.

Remarkably, we were unable to identify any qualitative or quantitative serological parameter predicting the occurrence of AMR. Recipients with AMR did not differ from nonrejecting patients with respect to DSA frequency, specificity or binding intensity. In this respect, desensitized transplant recipients may substantially differ from nondesensitized patients, where DSA detection was shown to predict AMR and inferior graft survival (18,47–49).

Notably, our regimen did not include adjunctive treatment with IVIG, rituximab or other antihumoral measures. It remains to be established, if by addition of such measures, a further improvement of outcomes can be achieved. There may be place for modifications of our protocol. For example, two highly sensitized recipients in urgent need of an allograft (not included in the present analysis) were registered for high urgency and upon registration subjected to an extended course of pretransplant IA until they were offered an allograft 1–2 weeks later. One recipient had a positive current CDCXM that could be rendered negative by the last pretransplant IA session. Transplantation was successful in both cases with excellent long-term outcome (data not shown).

In summary, we herein demonstrate that peritransplant IA enables successful DSA-positive deceased donor kidney transplantation. However, regarding the still increased rejection rates both among DSA-positive and DSA-negative recipients, we did not succeed in identifying any predictive serological parameter. More accurate diagnostic strategies for risk stratification remain to be established, and future studies will have to clarify if modifications regarding allocation to treatment or the addition of adjunctive treatment could further improve transplant outcomes.


The authors thank Romana Raab and Bettina Haidbauer for excellent technical assistance.

Funding sources: This study was supported by a grant from the Else-Kröner-Fresenius Stiftung (to G.A.B and G.B.; project number: P89/08-A117/08).


The authors of this manuscript have no conflicts of interest to disclose as described by the American Journal of Transplantation.