Journal Logo


Cold Ischemia is a Major Determinant of Acute Rejection and Renal Graft Survival in the Modern Era of Immunosuppression

Mikhalski, Dimitri1; Wissing, Karl Martin2; Ghisdal, Lidia2; Broeders, Nilufer2; Touly, Marie2; Hoang, Anh-Dung1; Loi, Patricia1; Mboti, Freddy1; Donckier, Vincent1; Vereerstraeten, Pierre2; Abramowicz, Daniel2,3

Author Information
doi: 10.1097/TP.0b013e318169c29e
  • Free


In renal transplantation, two of the main adverse early posttransplant events that one would like to avoid are rejection episodes and delayed graft function (DGF). Kidney graft function after transplantation is a continuum that ranges from immediate normal function to absence of function and anuria. DGF commonly refers to the need for posttransplantation dialysis during the first postoperative week. The rate of DGF after cadaveric transplantation range between 15% and 30% (1–4). DGF is a major adverse event as it makes the diagnosis of rejection difficult, it prolongs hospitalization and in some, but not all studies, is associated with an increased incidence of rejection and a reduced graft survival (5–7). The main risk factors for DGF are increasing donor age, longer cold ischemia time (CIT), and presence of anti-human leukocyte antigen (HLA) antibodies (8).

Several important changes have taken place in transplantation medicine in the last decade. First, the age of both donors and recipients has increased steadily, and this might impact on DGF rates. Second, the widespread use of mycophenolate mofetil, tacrolimus, and anti-interleukin-2 receptor monoclonal antibodies have resulted in a large reduction of acute rejection (AR) rates, from 50% a decade ago to 10% to 15% nowadays. Finally, with the widespread use of tacrolimus, it is of interest to compare DGF rates between patients receiving either tacrolimus or CsA at day 0 of transplantation. Therefore, we believed it would be worthwhile to revisit the relationships between CIT, the use of a calcineurin inhibitor (CNI) on day 0, DGF, AR rate, and long-term graft survival in a recent cohort of renal transplant recipients.

We have conducted a single-center retrospective study of a cohort of 611 patients, who have been transplanted between 1996 and 2005 to try to shed some light on three main questions: (1) What are the risk factors for DGF in our patients? (2) Are CIT and/or DGF independent risk factors for AR episodes? (3) Are CIT and/or DGF independent risk factors for long-term graft loss?



From January 1, 1996 to October 22, 2005, 564 patients underwent 714 renal transplantations at our institution. From those, we excluded from the study 44 pediatric transplants (two with combined liver transplantation), and 41 combined transplantations performed on adults (29 pancreas, 8 livers, and 4 hearts). As our aim was to study the long-term impact of CI and DGF on graft survival, we also excluded 18 transplants with technical failure during the first postoperative month (seven arterial and/or venous thromboses, four urological problems, and seven primary nonfunction), thus leaving 611 transplants performed on 475 patients. There were 43 transplants performed with living donors, and 568 with cadaveric donors. There were no donors after cardiac death in this cohort. The median follow-up was 62 months (interquartile range, 36–95 months).

DGF was defined as the need for dialysis in the postoperative period. The duration of DGF was defined as the number of days between the date of grafting and the last hemodialysis required before serum creatinine spontaneously decreased. Patients with DGF were routinely biopsied on a weekly basis to be able to diagnose concomitant AR on histological grounds.

Most patients (581 of 611; 95%) received a CNI as maintenance therapy, either cyclosporine (n=263) or tacrolimus (n=318), and corticosteroids in tapered doses. There was no systematic policy of CNI dose reduction in case of DGF. Most of them (535 of 611; 87.6%) received also either azathioprine (n=138) or mycophenolate mofetil (n=397). Thirty patients (4.9%) were given sirolimus with no CNI at the start of transplantation. Four hundred sixty-seven of the 611 grafts were given an induction with an antilymphocyte preparation [antithymoglobulin (n=106), OKT3 monoclonal antibody (n=140), an antagonist of the IL2-receptor (n=213), and anti-LFA-1 (human leukocyte-function-associated antigen-1) monoclonal antibody (n=14)]. The introduction of the CNI was delayed by several days in patients who received either antithymocyte globulin or OKT3 monoclonal antibody (n=246). Altogether, 335 patients (54.8%) did receive a CNI at day 0, and 276 (45.2%) did not.

High immunological risk patients were defined as those with either current panel-reactive lymphocytotoxic antibody (PRA) more than 5% at the time of the graft, peak PRA more than 50%, rapid (<2 years) immunological loss of a previous kidney graft, and those receiving a third or a fourth renal transplantation.

Before transplantation, patients received isotonic saline in a volume approximately equivalent to 3% of their ideal body weight, either during the pretransplant dialysis or before transplantation, through a peripheral vein, at a maximal flow of 1 L/hr. During the surgical intervention, the central venous pressure was targeted between 6 and 12 cm H2O through the infusion of isotonic saline. After surgery, in the recovery room, if diuresis was above 20 mL/hr, urinary losses were compensated volume by volume, at a maximal rate of 200 mL/hr for the initial eight postoperative hours. If diuresis was below 20 mL/hr, then central venous pressure was checked, and again brought between 5 and 10 cm H2O; if diuresis did not increase, patients received furosemide intravenously, 250 mg/6 hr.

Statistical Methods

Differences between nominal and continuous variables were tested by Fisher's or chi-square tests (when more than two groups were compared) and by analysis of variance with a posteriori Bonferroni-Dunn tests, respectively.

Logistic regression analysis was used to test the relationship between several independent (nominal or continuous) variables and a dependent nominal variable.

Univariate survival studies were conducted according to the Kaplan–Meier method, and Breslow–Gehan–Wilcoxon test (which takes into account the patients at risk at each point of the survival curves) was used to evaluate differences between the survival curves.

Cox's proportional hazard method was used in multivariate survival studies. Covariates were stepwise entered into the model to select significant (P<0.05) prognostic factors. Patient's death was censored from the analysis of graft survival.

The following covariates were considered in logistic regression analyses and survival studies: donor's origin (cadaver vs. living), recipient's and donor's sex and age (years), year of transplantation, number of the graft (first vs. regraft), number of pretransplant blood transfusions, duration of the pretransplant dialysis period (years), warm (min) and cold (h) ischemia times, high versus low immunological risk patients, number of HLA-A, B and DR mismatches, DGF (absent vs. present) and immunosuppression stratified according to the protocol that was used on an intention to treat basis.


Delayed Graft Function

DGF was observed in 99 of the 611 transplants under study (16.2%) with a mean duration of 8.5 days (median, 6; range, 1–40 days) and lasted less than 1 week in 57% and less than 2 weeks in 85% of grafts. Demographic characteristics of the two groups are reported in Table 1. Univariate analyses show that patients with DGF were more often transplanted before the year 2000, had high immunological risk, had longer cold ischemia, longer duration of pretransplant dialysis, and more have had a previous graft when compared with those free of DGF. The odds ratio of DGF was 1.08 (95% CI 1.05–1.12) for each 1-hr increase in cold ischemia. DGF occurred in 16.1% of patients who received a CNI at day 0, when compared with 16.3% among those who did not (P=NS). The incidence of DGF was similar whether patients received CsA or tacrolimus (17.4% vs. 16.1%, P=NS). In our cohort, the age of the donor had no impact on DGF rate by univariate analysis. However, it must be noticed that old donors were more prevalent among all categories associated with lower risk of DGF (grafts after year 2000, short cold ischemia, short dialysis time, higher proportion of first tranplants). When these confounding factors were taken into account in the multivariate analysis, donor age became significantly associated with the risk of DGF. CIT was the most important independent predictor of DGF (P<0.001). Using the constant and the coefficients of the logistic regression equation (ln P/(1−P)=−3.192+0.079×cold ischemia), the risk of DGF can be predicted for each value of cold ischemia. As examples, the risk of DGF for cold ischemia increasing from 12, 18, 24, 30, and 36 hr are 9.6%, 14.6%, 21.5%, 30.5%, and 41.4%, respectively. When we include in the analysis 14 patients who failed to establish graft function (seven with primary nonfunction and seven with graft vessels thromboses), the rate of DGF increased from 16.2% to 18.1% (113 of 625 patients). The inclusion of these 14 patients did not modify the results of the risk factors for DGF presented in Table 1: by multivariable analysis, donor age (P=0.005) and CIT (P<0.001) remained the two independent predictors of DGF.

Patient demographics and risk factors for delayed graft function

Relationship Between DGF and First-Year Acute Rejection

During the first postoperative year, rejection occurred in 16.5% of grafts (n=101). Twenty grafts (3.3%) presented two rejection episodes. Twelve grafts were lost during the first year because of irreversible rejection (2%).

Demographic characteristics of grafts with and without first-year AR are shown in Table 2. Univariate analyses show that significant factors of AR are recipient's age (increase in younger patients), year of transplantation (decrease during the last 6 years), a higher number of pretransplant blood transfusions, longer duration of pretransplant dialysis, longer CIT, high immunological risk, number of HLA mismatches and presence of DGF. By univariate analysis, the proportion of patients who experienced first-year rejection was higher among those with DGF (24 of 101; 24%) when compared with those free of DGF (77 of 510, 15%) (P=0.037). Along the same line, although the number was small, the proportion of patients who experienced irreversible rejection was higher among those with DGF (5 of 101, 5%) when compared with those free of DGF (7 of 510; 1.4%) (P=0.034). The proportion of patients with rejection did not differ significantly between the various immunosuppressive protocols (not shown).

Patient demographics and risk factors for acute rejection

With regard to the 24 patients who experienced both DGF and AR, AR occurred during the period of DGF for 14 of them. In these 14 patients, AR occurred at day (mean±SD) 6.1±3.8, and DGF lasted 16±13 days. For the remaining 10 patients, DGF lasted 5.2±2.8 days, and rejection took place at day 66±81.

Five factors emerged as independent predictors of rejection in the multivariate logistic regression analysis: the year of transplantation (less rejection in the most recent era), longer duration of dialysis, longer CIT, current PRA more than 5%, and the number of HLA-A, B and DR mismatches (Table 2). Thus, DGF in our cohort is not an independent risk factor for rejection. After adjustment for the other risk factors CIT was associated with a 4% increase in the odds of AR per 1 hr of CIT (OR 1.04; 95% CI 1.003–1.07; P=0.03). Accordingly, the observed incidence of first-year AR was 14.1% in patients with less than or equal to 24 hr CIT versus 29.3% in patients with more than 24 hr CIT, P<0.001.

Impact of DGF, CIT, and AR on Long-Term Survival

Patient survival was 94.2% (95% CI 92%–96%) at 5 years. Older recipient age (hazards ratio [HR] 1.05 per 1-year increase in age; 95% CI 1.02–1.08; P<0.001) was the only significant predictor of patient death among the variables studied.

The effect of DGF and other factors on death-censored graft survival was assessed by univariate and multivariate analyses. By univariate analysis, DGF was associated with a trend to a reduced graft survival. At 5 years, graft survival was 90.8% (95% CI 87.6%–93.1%) among patients free of DGF, vs. 85.9% (95% CI 76.9%–91.6%) among patients with DGF, P=0.06 (Fig. 1a). The detrimental effect of DGF on graft survival was entirely explained by the increased incidence of AR in patients with DGF. Indeed, as shown in Figure 1(b), after stratification for AR, we did not observe a difference in graft survival among patients with and without DGF. Cold ischemia of more than 24 hr was also associated with a trend to increased death-censored graft loss when compared with shorter CITs (Fig. 2a; P=0.09). Again, this increase in graft loss was totally explained by the increased risk of AR in patients with prolonged CIT. As shown in Figure 2, stratification for AR abolished the differences in graft survival in patients with CIT of more and less than 24 hr. The most potent predictors of death-censored graft loss in the univariate Cox models are first-year AR (HR 5.22; P<0.001), recipient age (HR 0.96 per 1-year increase, i.e., lower risk in older patients, P<0.001), retransplantation (HR 2.5; P<0.001), historical PRA more than 5% (HR 2.5; P<0.001), current PRA more than 5% (HR 2.1; P=0.01), and CIT (HR 1.04 for each 1-hr increase in cold ischemia; P=0.035).

A, Death-censored graft survival in patients with and without DGF. B, No impact of DGF on graft survival after stratification for acute rejection.
A, Impact of cold ischemia time on death-censored graft survival. B, No impact of cold ischemia on graft survival after stratification for acute rejection.

For the multivariable Cox model, we took two approaches, the first one by considering only the pretransplant variables (thus leaving out first-year AR and DGF). In that setting, younger recipient age, historical PRA (>5%), and longer cold ischemia remained independently associated with death-censored graft loss (Table 3). When first-year AR and DGF were included in the model, then the occurrence of first-year rejection, younger recipient age, and retransplantation were significant predictors of graft loss. Therefore, DGF had no significant association with death-censored graft loss neither in the univariate nor in the multivariate Cox models.

Multivariable analysis of death-censored graft survival


The first finding from our study is that DGF incidence was 16%, which is at the lower end of the reported frequencies after cadaveric donor transplantation. Indeed, data from UNOS report an incidence of DGF of 22% to 25% for cadaveric donors which has not changed between 1991 and 2003 (1). Along this line, recent single-center (3) and multicenter studies have reported DGF rates ranging from 20% to 30% (2, 4). Our low DGF rate might be a result of the aggressive patient hydration before transplantation (9–11).

DGF rates were independent of the use of CNI pretransplantation. This is in line with recent studies where DGF rates where similar, approximately 20% to 25%, whether patients received a CNI pretransplant or whether it was delayed by several days (2, 12, 13). As expected, rates of DGF were similar whether the CNI used pretransplant was CsA or tacrolimus.

Risk factors for DGF have been identified in large multicenter trials and can be ascribed to donor factors, recipient factors, or CIT. The most powerful and robust factors are with regard to the donor: donor age, non-heart-beating donors, history of donor hypertension, and terminal donor creatinine; with regard to the recipient: HLA sensitization, number of HLA mismatches, previous transplantation, and male sex; and the duration of cold ischemia (5, 8). Consistent with these registries, we found in our database that high immunological risk patients showed higher rates of DGF, whereas donor age and cold ischemia were independent predictors of DGF by multivariate analysis. We next evaluated the relationship between cold ischemia, DGF, and first-year AR incidence, which occurred in 16% of our patient population. Of note, this low rejection rate indeed confirms that our patient population belongs to the “modern era” of immunosuppression. As expected, we found that HLA sensitization, number of HLA mismatches, and transplantation before 2000 were independent risk factors for rejection episodes. By univariate analysis AR were more frequent among patients who experienced DGF and among those with prolonged cold ischemia, but multivariable analysis revealed that only cold ischemia, and not DGF, was an independent predictor of AR. The association between DGF and AR is controversial, with some reports showing a significant association (3, 14), whereas other do not (1, 15, 16). With regard to cold ischemia, a recent registry analysis showed an association between CIT and AR incidence (17). In fact, the pathophysiologies of cold ischemia-induced DGF and rejection are very similar and have similar kinetics (18). After ischemia, graft reperfusion is associated with local chemokines and cytokines production, leading to recruitment of neutrophils, monocytes, and T cells into the graft. Acute tubular necrosis and DGF results from impaired capillary perfusion caused by leukocyte plugging, as well as from direct tubular toxicity from oxygen radicals and neutrophils proteases. At the same time, graft immunogenicity is increased because of enhanced expression of adhesion molecules such as ICAM-1 and VCAM-1, MHC class I and class II antigens, as well as upregulation of key T-cell costimulatory molecules such as CD80 and CD86. Therefore, there is a large body of pathophysiological evidence linking stepwise increase in CIT with higher rates of both DGF and AR.

The fact that some studies identified DGF as a risk factor for AR might be owing to several reasons. First, kidney graft rejection is more difficult to diagnose in the absence of graft function. It has therefore become routine practice in most units, including ours, to perform sequential biopsies in this setting to diagnose rejection on histological grounds during the last decade (19, 20). Second, the histological lesions of acute tubular necrosis, with its associated leukocyte infiltration, are sometimes difficult to differentiate from graft rejection. The introduction of the Banff scheme for acute renal allograft rejection has allowed for a better accuracy in the diagnosis of graft rejection in the setting of acute tubular necrosis. Third, in some centers there is a policy of CNI dose reduction in case of DGF in the hope to limit its severity, a strategy that is likely to increase the risk of rejection (21). Finally, HLA sensitization and previous transplantation are risk factors for DGF, probably because those patients may develop very rapid AR that contributes to DGF occurrence. This might further blur the causal association between DGF and AR. A more plausible hypothesis is that there exists a causal relationship between early AR and DGF, rather than the reverse.

When turning to factors affecting long-term graft survival, in our series the most important predictor of graft loss was the occurrence of first year AR. Indeed, graft loss was 4.6 times more frequent in those patients when compared with rejection-free patients. In contrast to previous observations, (1, 3, 14, 16, 20), we did not observe worse graft survival in those patients who experienced both DGF and AR, and likewise, DGF had no impact on GS. Although the reasons for these discrepant results are not clear, it is worth noting that data showing an association between DGF and either rejection incidence or long-term graft survival were obtained during the previous era of immunosuppression, as shown by rejection rates ranging from 25% to 55% (1, 3, 14, 16, 20). In addition, one may argue that underdiagnosis of rejection in case of DGF was more prevalent in these earlier cohorts, as discussed above. On the other hand, in our series prolonged CIT was a significant independent risk factor for graft loss, but this effect was entirely because of its impact on rejection. Indeed, in the absence of AR, patients with CIT more than 24 hr had 10-year graft survival rates that were virtually identical to patients with CIT shorter than 24 hr.

What are the possible implications of these data in term of initial patient management? Obviously, efforts should be attempted to limit cold ischemia to a reasonable duration. Data from both USA and Europe show stepwise yearly improvements in that regard (1, 17). Second, peritransplant volemia should be optimized, as it is a critical modifiable factor that influences the occurrence of DGF (9–11). Our data also suggest that special attention should be given to the immunosuppressive therapy for patients who receive a graft with prolonged ischemia times. Indeed, these patients are at increased risk of rejection, and should be given efficient rejection prophylaxis. In that respect, we showed a decade ago that, for patients with CIT above 24 hr, prophylaxis with the OKT3 monoclonal antibody was associated with significantly less AR and improved 5-year graft survival when compared with patient not given OKT3 induction (22). What are the options today, where OKT3 is not in use any more? Data from a recent trial where patients at high risk for DGF were randomized to receive either antithymocyte globulin (ATG) or basilixumab showed similar DGF rates (∼40%), but significantly less rejection episodes in the ATG group (23). Therefore, we believe that patients with prolonged CIT, rather than those who go on to develop DGF, should receive antibody induction therapy. The choice between an anti-IL2 receptor monoclonal antibody and ATG may depend on the presence of other immunological risk factors for rejection that, if present, would in our view support the use of ATG.


The authors acknowledge the important contribution of Brigitte Borre, Francoise Bernard, and Sylvie Arias Lopez, our clinical research nurses to this work.


1. Cecka JM. The OPTN/UNOS renal transplant registry 2003. Clin Transpl 2003; 1–12: 1.
2. Kamar N, Garrigue V, Karras A, et al. Impact of early or delayed cyclosporine on delayed graft function in renal transplant recipients: A randomized, multicenter study. Am J Transplant 2006; 6: 1042.
3. Quiroga I, McShane P, Koo DD, et al. Major effects of delayed graft function and cold ischaemia time on renal allograft survival. Nephrol Dial Transplant 2006; 21: 1689.
4. Silva LA, Felipe CR, Park SI, et al. Impact of initial exposure to calcineurin inhibitors on kidney graft function of patients at high risk to develop delayed graft function. Braz J Med Biol Res 2006; 39: 43.
5. Lebranchu Y, Halimi JM, Bock A, et al. Delayed graft function: Risk factors, consequences and parameters affecting outcome-results from MOST, A Multinational Observational Study. Transplant Proc 2005; 37: 345.
6. Rosenthal JT, Danovitch GM, Wilkinson A, et al. The high cost of delayed graft function in cadaveric renal transplantation. Transplantation 1991; 51: 1115.
7. Sanfilippo F, Vaughn WK, Spees EK, et al. The detrimental effects of delayed graft function in cadaver donor renal transplantation. Transplantation 1984; 38: 643.
8. Irish WD, McCollum DA, Tesi RJ, et al. Nomogram for predicting the likelihood of delayed graft function in adult cadaveric renal transplant recipients. J Am Soc Nephrol 2003; 14: 2967.
9. Abramowicz D, De Pauw L, Le Moine A, et al. Prevention of OKT3 nephrotoxicity after kidney transplantation. Kidney Int Suppl 1996; 53: S39.
10. Schnuelle P, Johannes van der WF. Perioperative fluid management in renal transplantation: A narrative review of the literature. Transpl Int 2006; 19: 947.
11. Snoeijs MG, Wiermans B, Christiaans MH, et al. Recipient hemodynamics during non-heart-beating donor kidney transplantation are major predictors of primary nonfunction. Am J Transplant 2007; 7: 1158.
12. Charpentier B, Rostaing L, Berthoux F, et al. A three-arm study comparing immediate tacrolimus therapy with antithymocyte globulin induction therapy followed by tacrolimus or cyclosporine A in adult renal transplant recipients. Transplantation 2003; 75: 844.
13. Mourad G, Garrigue V, Squifflet JP, et al. Induction versus noninduction in renal transplant recipients with tacrolimus-based immunosuppression. Transplantation 2001; 72: 1050.
14. Tejani AH, Sullivan EK, Alexander SR, et al. Predictive factors for delayed graft function (DGF) and its impact on renal graft survival in children: A report of the North American Pediatric Renal Transplant Cooperative Study (NAPRTCS). Pediatr Transplant 1999; 3: 293.
15. Pieringer H, Biesenbach G. Risk factors for delayed kidney function and impact of delayed function on patient and graft survival in adult graft recipients. Clin Transplant 2005; 19: 391.
16. Troppmann C, Gillingham KJ, Benedetti E, et al. Delayed graft function, acute rejection, and outcome after cadaver renal transplantation. The multivariate analysis. Transplantation 1995; 59: 962.
17. Opelz G, Dohler B. Multicenter analysis of kidney preservation. Transplantation 2007; 83: 247.
18. Perico N, Cattaneo D, Sayegh MH, et al. Delayed graft function in kidney transplantation. Lancet 2004; 364: 1814.
19. Gaber LW, Gaber AO, Hathaway DK, et al. Routine early biopsy of allografts with delayed function: Correlation of histopathology and transplant outcome. Clin Transplant 1996; 10: 629.
20. Howard RJ, Pfaff WW, Brunson ME, et al. Delayed graft function is associated with an increased incidence of occult rejection and results in poorer graft survival. Transplant Proc 1993; 25: 884.
21. Thervet E, Pfeffer P, Scolari MP, et al. Clinical outcomes during the first three months posttransplant in renal allograft recipients managed by C2 monitoring of cyclosporine microemulsion. Transplantation 2003; 76: 903.
22. Abramowicz D, Norman DJ, Vereerstraeten P, et al. OKT3 prophylaxis in renal grafts with prolonged cold ischemia times: Association with improvement in long-term survival. Kidney Int 1996; 49: 768.
23. Brennan DC, Daller JA, Lake KD, et al. Rabbit antithymocyte globulin versus basiliximab in renal transplantation. N Engl J Med 2006; 355: 1967.

Acute rejection; Calcineurin inhibitor; Cold ischemia time; Delayed graft function; Graft survival; Kidney transplantation

© 2008 Lippincott Williams & Wilkins, Inc.