The long-term graft outcomes after positive cross-match (PXM) living donor kidney transplantation (LDKT) are unknown and the descriptive published data present short-medium term results. We conducted a retrospective cohort study of LDKT with PXM by flow cytometry performed at our center during February 1999 to October 2006, compared to a control group, matched 1:1 for age, sex, race, retransplantation and transplant year. The PXM group was treated with a course of plasmapheresis/low-dose intravenous immunoglobulin (IVIg) preoperatively, and OKT3 or thymoglobulin induction.
Both groups (n = 41 each) were comparable except for duration of end-stage renal disease (ESRD), induction, HLA mismatch and panel-reactive antibody (PRA). During the period of up to 9 years, 14 PXM and 7 controls lost their grafts (p < 0.04). Graft survival rates at 1 and 5 years were 89.9% and 69.4% for PXM group and 97.6% and 80.6% for the controls, respectively. PXM was associated with higher risk of graft loss (HR 2.6, p = 0.04; 95%CI 1.03–6.4) (t1/2= 6.8 years), but not with patient survival (HR 1.96, p = 0.29; 95%CI 0.6–7.0) or 1-year serum creatinine (β= 0.06, p = 0.59 for ln (SCr); 95% CI −0.16 to 0.28).
These results suggest that despite the favorable short-term results of PXM LDKT after PP/IVIg conditioning, medium-long-term outcomes are notably worse than expected, perhaps comparable to non-ECD deceased donor kidney transplantation (KT).
The role of anti-donor HLA antibody in causing hyperacute rejection and graft loss was recognized in early years of clinical transplantation (1,2). With introduction of pretransplant cross-match (XM) with donor lymphocytes (3,4), this catastrophic event was to a large extent prevented. With development of more sensitive and specific techniques, we have been able to identify a larger number of candidates who have HLA antibodies from previous transplantation, transfusion or pregnancy. Currently, 34.4% of the waitlisted patients in the Unite States are sensitized to HLA antigens; 14.4% with peak percentage of panel-reactive antibody (PRA) >80 (5). These patients are less likely to find compatible deceased donors (DD) and have a significantly longer waiting time. With the lack of compensated unrelated living donation in the Unite States (6) and limited possibility of paired kidney exchange (7,8), these patients have no other alternative for transplantation. The initial reports from our group (9) and others (10) demonstrated the feasibility of successful positive (P)XM living donor kidney transplantation (LDKT). With increasing experience over the past decade, different centers have published their short-medium term results using various desensitization protocols (11,12). The ongoing advancements in the XM techniques and methods of donor-specific antibody (DSA) detection over the past decade and the changes in treatment protocols have made the report of these results more challenging. Despite the progress, we are still far from certainty in confirming the presence or absence of all forms of DSA in the recipients.
In spite of a number of observational studies on short-term outcomes, the long-term results of PXM LDKT and its comparison with negative XM transplantation have never been reported. Examining the true impact of desensitization on graft outcomes with randomized, prospective trials would not be feasible and prospective cohort studies would take a long time to complete. Since this information is crucial for the health care system to endorse the available options for the growing sensitized patient population, we sought to evaluate the medium-long-term graft outcomes after conditioning in PXM LDKT recipients.
In this retrospective, comparative cohort study we examined the long-term results of LDKT after desensitization therapy for PXM by flow cytometry (FC). All LDKTs performed at our center after desensitization between February 1999 and October 2006 were reviewed. We blindly selected a control group from a pool of LDKT recipients with negative XM transplanted during the same period and were matched 1:1 for gender, race, age, previous KT and year of transplantation. All renal allografts were retrieved by laparascopic method (13).
HLA typing, and cross-match and antibody screening methods
As with other centers, the histocompatibility techniques used for HLA typing, XM and detection of anti-HLA DSA in our center evolved during the study period (14–20). A three-color FCXM was used to evaluate the presence of alloreactive antibody to donor T- and/or B-lymphocytes for all patients in this cohort. A XM was deemed positive for initiation of plasma exchange (PE)/low dose (LD) intravenous immunoglobulin (IVIg) if the shift median channel value (sMCV) from the negative control serum for the assay was greater than three standard deviations from the mean. This cutoff value was determined following statistical analysis of median channel values from at least 100 different individuals in cross-matches with a serum known to be antibody negative. Because of the changes in the instruments over the study period and batch-to-batch variability of the reagents, we did not use a single numeric cutoff value. To detect anti-HLA DSA, over these years we moved from anti-human globulin complement-dependent cytotoxicity (AHG-CDC) to ELISA and more recently FC assay using single-antigen beads. HLA typing was performed using low-resolution molecular methods for the majority of the cases. High-resolution allele-level typing was performed only in a minority of donors.
Desensitization and immunosuppressive regimen
Transplant candidates with positive XM results underwent a desensitization protocol that consisted of a course of alternate-day total PE followed by infusion of LD IVIg (100 mg/kg) (Pentaglobulin, Biotest, Othmarsingen, Switzerland, or Gammagard Liquid, Baxter Healthcare Corporation, Westlake Village, CA). Plasma volume was replaced with 5% albumin solution; fresh frozen plasma was administered on the preoperative day. After each treatment XM was repeated and if the FCXM median channel shift was reduced to within three standard deviations from the mean, transplant surgery was scheduled for the following day. Patients did not have additional PE/LD IVIg after surgery and did not receive rituximab (Genentech, San Fransisco, CA) unless they developed biopsy-proven acute antibody-mediated rejection (AAMR). All patients received antibody induction therapy. Those transplanted prior to 2002 received a 7-day course of muronomab OKT3 (Orthoclone, JANSSEN-CILAG) 5 mg/day. The majority of cases transplanted afterwards received a 7-day course of thymoglobulin (Genzyme, Cambridge, MA) (rATG) 1.5 mg/kg, initiated intraoperatively. Dosing was adjusted for leukopenia or thrombocytopenia with target CD3 count <50 cells/mm3. Only one patient was treated with two doses of 20 mg basiliximab (Simulect, Norvartis Pharmaceuticals, NJ). Two weeks prior to the anticipated transplant day patients were initiated on tacrolimus (Prograf, Astellas, Deerfield, IL) (TAC) 0.1 mg/kg/day with target level 10–15 ng/mL (by immunoassay), and mycophenolate mofetil (CellCept, Roche, Nutley, NJ) (MMF) 2 g/day, continued following transplantation with dose adjustment for side effects. Following an intraoperative dose of intravenous methylprednisolone 500 mg, additional daily doses of 250 and 125 mg were administered and a tapering dose of oral prednisone was initiated. The daily dose was tapered to 10 mg by 3 months and 5 mg by 1 year.
Recipients in the control group received similar maintenance regimen. However, less than half of them did not receive antibody induction therapy according to the institutional protocol during the early study period.
Patients with delayed graft function (DGF), defined as need for dialysis during the first week after surgery, and slow graft function (SGF), defined as serum creatinine (SCr) >3.0 mg/dL on day 5 posttransplant, underwent allograft biopsies every 7–10 days until SCr was stabilized. Protocol biopsies on predefined time points were not performed. After initial hospital stay, patients were followed as outpatient according to the recommended guidelines (21). An unexplained increase in SCr was followed by transplant ultrasound and biopsy, if indicated. Biopsies were evaluated using Banff 97 criteria (22,23) for evidence of acute cellular rejection (ACR), AAMR, transplant glomerulopathy (TG), thrombotic microangiopathy (TMA) and BK nephropathy. Patients with AAMR were treated with intravenous steroid, a thrice weekly course of PE/LD IVIg, and in three cases rituximab in addition to optimization of maintenance regimen. Follow-up biopsies were performed as needed to evaluate response to therapy.
The primary outcome of the study was graft survival. Graft loss included graft failure and death with functioning graft. The secondary outcomes included patient survival, SCr at various defined time points, ACR or AAMR and BK nephropathy. Data are presented as mean ± SD or counts and percentages, as appropriate. Comparison between the groups regarding SCr at 12, 36 and 60 months, and rate of acute rejection was made using Student's t-test, and chi-square and Fisher's exact test, as appropriate. Survival analysis methods including Kaplan–Meier estimates and Cox proportional hazard models were used to evaluate the effect of XM positivity, and the XM results after desensitization on graft and patient survival. To examine the independent association between presensitization and 1-year SCr, linear regression models were used with adjustment for potential confounding variables. Logarithmic transformation was utilized to normalize the distribution of outcome variable. In the multivariate models variables with p-value <0.1 were included. Model assumptions for regression models were tested using appropriate statistical diagnostics. Stata/SE 10.0 (StataCorp, College Station, TX) was used for statistical analysis.
During the study period, 41 patients with PXM by FC underwent LDKT at our center after desensitization. Patient and transplant related characteristics are summarized in Table 1. As expected, recipients with PXM had longer duration of ESRD, higher PRA and more degree of HLA mismatch with their donors. In contrast to patients in control group, all PXM patients received induction therapy.
Table 1. Patient and transplant-related characteristics
Positive XM group (n = 41)
Control group (n = 41)
1p = 0.03.
*p < 0.002.
42.8 ± 12.2
42.8 ± 12.4
27.8 ± 6.9
26.9 ± 7.1
Cause of ESRD
ESRD duration1 (year)
4.6 ± 4.5
1.8 ± 2.3
54.7 ± 30.1
15.6 ± 27.4
43.0 ± 33.9
9.3 ± 21.9
3.8 ± 1.4
2.7 ± 1.6
3.9 ± 2.2
5.2 ± 2.0
39.4 ± 10.7
41.5 ± 10.7
27.9 ± 4.9
26.5 ± 4.2
0.93 ± 0.14
0.91 ± 0.18
In the study group prior to desensitization, 33 had positive T-FCXM, and 35 were positive with B-FCXM. Twenty seven patients were both T- and B- FCXM positive. These patients underwent 1–12 sessions of PE/LD IVIG treatment prior to surgery (3.6 ± 2.2). Only three received more than six treatments. After desensitization therapy and prior to operation, 18 patients remained T-FCXM positive and 17 had positive B-FCXM, despite reduced sMCV; 15 with both tests positive. HLA Class I DSA was identified in 14 patients and class II DSA in five.
One PXM patient and three controls had DGF; two PXM and four control cases had SGF and one in the PXM group experienced primary nonfunction due to AAMR. Patients were followed for up to 9 years (3.9 ± 2.2 and 5.2 ± 2.0 years, respectively, p = 0.8) until their last visit or graft loss. During the study period, 21 patients in the PXM and 19 in the control group underwent clinically indicated biopsies. The median number of biopsies was 1 (range: 1–10) and 1 (range 1–5), respectively. Five PXM and six control patients experienced at least one episode of ACR (p = 0.5). AAMR was diagnosed in five patients in the PXM group within the first 10 days after surgery compared to none in the control group (p < 0.03); two lost their grafts as a result. T- and B-cell XM at the time of surgery were negative in three, and two had prior transplants. Three presensitized and four nonsensitized recipients developed TG (p = 0.5); TMA was reported in two cases in each group. Two patients in the former and one in the latter group developed BK nephropathy.
Serum creatinine at various time points is summarized in Figure 1. There was no statistically significant difference between PXM group as compared to controls at any of these time points (1.7 ± 0.8 vs. 1.5 ± 0.6 at 1 year, p = 0.3; 1.5 ± 0.6 vs.1.7 ± 0.8, p = 0.5 at 3 years; and 3.4 ± 2.4 vs. 2.0 ± 1.2, p = 0.09 at 5 years, respectively). Furthermore, XM positivity was not associated with 1-year SCr after adjusting for ACR and HLA MM, the only significant predictors in univariate analysis (β= 0.06, p = 0.59 for ln [SCr]; 95% CI −0.16 to 0.28).
During the study period, 14 presensitized and seven nonsensitized patients lost their grafts, five and four due to death with functioning graft, respectively. Figure 2 depicts the graft survival estimates for the two groups (p < 0.04). Graft survival at 1 year was 89.9% for the presensitized patients and 97.6% for the controls. Corresponding rates at 5 years were 69.4% and 80.6%, respectively. The actuarial graft t1/2 for PXM patients was 6.8 years. To evaluate the independent association between graft survival and PXM, we first examined its association with induction therapy and HLA MM that were different between the groups. Neither of the two factors were significantly associated with graft loss (for induction, HR 1.4, p = 0.66 for anti-CD25, and 2.7, p = 0.13 for lymphocyte-depleting agent, compared to no induction; and for each additional HLA MM HR 1.2, p = 0.12). After adjusting for ACR, presensitization was associated with inferior graft survival during the observation period (HR for graft loss: 2.6, p = 0.04; 95% CI 1.03–6.4). Similarly, AAMR was a strong predictor of worse graft survival in the PXM group (HR 9.1, p < 0.001, 95% CI 2.9–28.8). The 1- and 5-year graft survival rates were 90.6% and 69.2% for patients with positive T-cell FCXM and 87.5% and 72.9% with only B-cell FCXM positivity, respectively (p = 0.96 for graft survival comparison between the two). Overall, six patients in the PXM group and four in the control group died during this period, one in the former from sepsis. One PXM patient died from trauma, while two controls died because of gastrointestinal bleeding and pulmonary embolism. The causes of death in other patients were unclear. Comparing the two groups, patient survival was comparable (HR 1.96, p = 0.29; 95% CI 0.6–7.0).
To further identify the predictors of poor outcome within the PXM group, we examined the association of age, sex, race, BMI, pretransplant diabetes, prior transplantation, type of induction, early graft function and occurrence of AR with graft loss. Among these factors, only DGF/SGF (HR 26.1, p = 0.001; 95% CI 1.1–10.6), prior transplantation (HR 3.0, p < 0.05; 95% CI 1.0–8.7) and AR (HR 6.0, p = 0.001, 95% CI 2.0–17.9) appeared to be associated with graft loss. To evaluate the independent association between XM status after desensitization therapy and graft survival in those who were tested prior to operation, we entered these covariates in the regression model (Figure 3). Negative XM after PP/IVIg, prior to transplant surgery (n = 16) appeared to be associated with improved graft survival when compared with positivity in either or both T- and B-cell XM (n = 20), however, the association was not statistically significant (HR for graft loss: 0.45, p = 0.2; 95% CI 0.13–1.6). There was no significant association between results of FCXM after desensitization and ACR or AAMR after transplantation (data not shown). Furthermore, previous transplantation remained significantly associated with graft loss (HR 3.4, p < 0.05; 95% CI 1.0–11.8).
Within this group, when we compared graft survival in patients with DSA and those who did not have identifiable DSA with the method used at the time or were not tested, there was no difference (p = 0.44). Graft survival rates at 1 and 5 years were 83.3% and 70.5% in the former group and 90.5% and 68.3% for the latter, respectively.
In this study, we have presented the medium-long-term outcomes of LDKT in recipients with positive FCXM against donor cells following desensitization therapy. Moreover, in contrast to previous descriptive reports (11,12,24,25), we have compared these results to transplant outcomes in a group of matched recipients with negative XM. To decrease the risk of selection bias and the confounding effects, the control group was blindly chosen by 1:1 matching for age, gender, race, year of transplantation and retransplantation status from a pool of 946 LDKT patients transplanted at our center. We could identify only one matched control for some cases and for the others the control was chosen randomly. We acknowledge that we did not and could not match the two groups for all potentially important variables. To do so, one would need a very large pool of LDKT recipients, a remote possibility in a single center. Using available data bases would suffer other crucial shortcomings by ignoring factors like center effect and different maintenance immunosuppressive regimens.
The major differences between the groups in this cohort were the type of induction therapy, HLA MM, PRA and duration of dialysis. The latter three would be expected from the recipient's presensitization status. During the early years of the study period, antibody induction therapy was not utilized in the majority of recipients with negative XM. In more recent years, however, all recipients received induction therapy. In contrast, all patients in the study group received induction therapy, mostly OKT3 or rATG. Positive XM in our cohort was defined as positive T- and/or B-cell FCXM. This definition differs from that of Cedar-Sinai Medical Center (12) and Mayo Clinic (MC) (11). The former used T-cell CDC XM and the latter T-cell AHG XM. More recently, MC group has treated patients with positive FCXM. Our conditioning regimen included a course of pretransplant PE/LD IVIg, utilizing OKT3 or rATG for induction therapy. In contrast to the Johns Hopkins Hospital (JHH) (10) and MC (11) protocols, we did not administer rituximab (Genentech, CA) pretransplant and did not continue PE/IVIg after surgery. We believe the important contribution of this study is to present medium-long-term results of transplantation with PXM. Patients in this cohort were followed for up to 9 years. Despite the favorable short-term outcomes observed in this cohort and reported by others, the longer-term graft survival was worse than what would be expected from short-term results. Comparing graft survival in PXM recipients with matched controls showed that the former had a significantly worse graft survival, and XM positivity was found to be an independent predictor of poor transplant outcomes during the observation period. The rate of graft loss during the first year was higher (10.1% in PXM group vs. 2.4% in controls), partially due to higher number of failures from AAMR. The calculated t1/2 for graft survival in positive XM group was 6.8 years. The cause of late graft loss was death, sepsis and noncompliance, each in one case, and chronic rejection and AR, each in two.
Although the overall national 5-year graft survival rate for LDKT (79.7%) is considerably better (26), its comparison with this cohort consisting of 34% retransplant cases with high PRA and long duration of dialysis in addition to XM positivity could not be justified. Therefore, comparing the transplant outcomes in such cohort with a well-matched group of LDKT recipients with negative XM would be very informative. On the other hand, the corresponding graft survival rates for non-ECD and ECD DDKT are 70% and 53%, respectively (27). The former is comparable to 69.4% 5-year graft survival in our PXM cohort.
Sonnenday et al. (24) described the outcomes of LDKT in 18 patients treated with PE/LD CMVIg along with daclizumab (Roche, Nutley, NJ) induction. After a mean follow-up of 17.3 months (range 1–44) there was one reported graft loss. In a more recent review article, Montgomery et al. (28) reported an actuarial 3-year graft survival of just above 80% among more than 80 desensitized recipients.
Gloor et al. (11) in their first report described 79% short-term graft survival in 14 recipients using PE/LD IVIg/rituximab and splenectomy. Stegall et al. (25) subsequently described graft outcomes in 61 patients followed for 0.35–5.2 years. Thirty two patients received PE/LD IVIg/rituximab ± splenectomy; 13 were treated with high-dose (HD) IVIg (2–3 g/kg) and 16 with PP/LD IVIg/rituximab and intensive posttransplant DSA monitoring. The overall actuarial 1-year patient and graft survival was 93% and 82%, respectively. Among 10 patients who achieved low-titer but not completely negative XM, 50% lost their grafts and 70% experienced AAMR.
Higgins et al. (29) studied DSA levels after LDKT in 24 sensitized recipients, defined by CDC or FCXM or microbead assay. Patients were treated with a course of PE. In addition, two received rituximab and three IVIg preoperatively. Graft survival was 87.5% at 3 months.
West-Thielke et al. (30) recently reported their experience with 22 African-American and 28 non-African-American recipients with positive T- and B-cell FCXM who received PE/LD IVIg pre- and post-transplant ± rituximab. One-year graft survival rates were 82.6% and 94.1%, respectively. The actuarial 3-year graft survival was close to 65% in non-African-American and close to 55% in African-American group. Efficacy of immunoadsorption in 13 highly sensitized recipients of DD renal allografts with positive CDC and/or FCXM was demonstrated by Higgins and colleagues (31). During the median follow-up of 26 months, there were 6 graft losses and one death.
Glotz et al. (32) reported 15 patients with class 1 PRA > 50% who underwent desensitization with three monthly infusions of HD IVIg. Thirteen patients qualified for transplantation, 11 from DD and 2 from LD. After 12 months, patient and graft survival rates were 100% and 84.6%, respectively. Jordan et al. (12) subsequently described their experience with 42 patients with positive CDC XM who received HD IVIg. There were 26 patients who underwent LDKT and 16 who received DD organ transplants after four monthly infusions. The overall 2-year graft survival was 89.1%. Jordan and colleagues (33) conducted a prospective, randomized trial of four monthly infusions of HD IVIG in ESRD patients with PRA > 50%. Graft survival in 16 from the IVIG group (n = 46) who underwent transplantation was 80% after a median follow-up of 2 years. Vo et al. (34) recently reported the short-term results of subcutaneous alemtuzumab induction therapy along with HD IVIg and rituximab in 54 patients with positive CDC- or FC-XM with their donors. Graft survival rate at 1 year was 96%.
We observed five cases with AAMR shortly after surgery; two lost their grafts. The rates of ACR (5 cases) and TG (3 cases) were not different from the control group. We acknowledge that we did not perform protocol biopsies and may have missed subclinical pathologies. Moreover, during the course of this study pathological criteria used for diagnosis of AAMR evolved (22,23). Therefore, some cases could have been missed in earlier years. JHH group reported episodes of AAMR in five of 18 cases (24). Mayo Clinic group performed protocol biopsies in their cohort and observed AAMR in 80%, 37% and 29% of the patients, respectively, with three regimens described earlier (25). Higgins et al. (29) reported AAMR in 41.7% of 24 desensitized patients. West-Thielke and colleagues (30) observed AAMR in 35.7% of African-American and 27.3% of non-African-American recipients within 1 year posttransplant.
The SCr in PXM group at various time points after transplantation was comparable to the control group. Although there was an upward trend in the group SCr at 5 years, there was no significant increase in those who had serial levels available at the three time points. Sonnenday et al. (24) reported a mean SCr of 1.1 mg/dL after 17.3 months mean follow-up. In MC experience, the mean SCr at 6 months was 1.1, 1.8 and 1.6 mg/dL with their three protocols (25). Similarly, Jordan et al. (12) reported a mean SCr of 1.4 mg/dL at 2 years.
We further examined the predictive value of XM results after desensitization, at the time of surgery for graft survival. For patients with negative T- and B-cell XM graft survival was better than those who still had low-grade positivity before surgery, but the difference was not statistically significant. Similar to other reports (28), when we examined the role of previous transplantation, it was found to be an independent predictor of graft survival. Furthermore, analysis of repeat HLA mismatches in patients with previous grafts did not show any association with outcome (data not shown).
We are well aware of the limitations of this study. The matching variables were not exhaustive, which may raise concerns regarding internal validity of the study. To address this issue, we matched the groups a priori for a number of important variables. The groups were further comparable regarding a number of other factors and we tried to statistically adjust for the few discrepant potentially significant covariates. The relatively small number of cases limits the power of the study to detect true differences in the secondary endpoints and warrants cautious generalization of the results. Moreover, presence and type of DSA was not available in all cases. This shortcoming was unavoidable as the available techniques during the earlier years of the study were less sensitive than the current methods. Unfortunately, historical serum samples were not available for testing in those without identified DSA. This shortcoming also applies to all the other reports, which have relied on different XM methods instead of identification of DSA. With the recent methodological advances, recognition of anti-HLA antibodies has become possible to a large extent. However, these techniques still cannot identify DSA against uncommon HLA alleles, very low-titer DSA and non-HLA antibodies. Furthermore, detection of DSA using single-antigen microbeads in some cases is based on presence of antibody against a homologous HLA allele rather than the donor's allele. Therefore, XM results, even without identifiable DSA, are still crucial to identify this high-risk group. We also admit that the absence of scheduled protocol biopsies and posttransplant monitoring of DSA in this cohort could have adversely affected the outcomes.
Although we matched the control group for a number of important confounding variables, one cannot rule out the effect of other unaccounted confounders. We acknowledge that some other centers have performed a larger number of transplants after desensitization. However, to our knowledge this is the largest cohort with the longest follow-up so far published. In addition, to obtain a better understanding of the impact of XM positivity on graft outcomes, we compared this group with a well-matched group of LDKT recipients with negative XM.
In summary, the results of our study suggest that conditioning of the recipients with positive FCXM with their potential donors using pretransplant PE/LD IVIg can provide fair short-term results, but suboptimal medium-long-term outcomes, inferior to XM negative LDKT. The 5-year survival rates seem to be comparable to non-ECD and perhaps better than ECD DDKT. Achieving complete T- and B-cell FCXM negativity at the time of surgery is probably associated with better graft survival. Further studies are required to confirm our observations and to identify risk-stratified treatment options to reduce the risk of AAMR and graft loss due to antibody-mediated tissue injury.