Early Graft Function After Living Donor Kidney Transplantation Predicts Rejection But Not Outcomes


*Corresponding author: Sandy Feng, fengs@surgery.ucsf.edu


Poor early graft function (EGF) after deceased donor kidney transplantation (DDKT) has been intensely studied. Much less is known about poor EGF after living donor kidney transplantation (LDKT). Data were collected on 469 LDKTs performed between 1/1/97 and 12/31/01 to determine risk factors for and outcomes associated with poor EGF, defined as either delayed or slow graft function (DGF or SGF). The incidence of DGF and SGF were 4.7% and 10.7%, respectively. Diabetic etiology (OR 2.22; p = 0.021) and warm ischemia time (WIT) (OR 1.05 per min increment; p = 0.0025) emerged as independently associated with poor EGF. Neither functional graft survival nor 1-year graft function differed among the EGF groups. However, DGF and SGF strongly predisposed to acute rejection (AR), which compromised functional graft survival (p = 0.0007) and 1-year graft function. Therefore, we conclude that diabetic etiology of renal disease and WIT are the dominant risk factors for poor EGF after LDKT. Poor EGF did not directly compromise functional graft survival but strongly predisposed to AR. We suggest that immunosuppression should be intensified in the poor EGF setting to maximize LDKT longevity, as AR does impair functional graft survival.


In the United States, living donor kidney transplantation (LDKT) has increased substantially during the past decade and, as of 2001, has surpassed deceased donor kidney transplantation (DDKT). Many factors have likely contributed to this escalation, including the increasing severity of the organ shortage, excellent donor and recipient outcomes, a broader acceptance of unrelated donors, and technical advances such as laparoscopic donor nephrectomy. It is well known that LDKT enjoy superior outcomes compared with DDKT (1–3). The differences have been predominantly attributed to both superior organ quality – a normal kidney from a healthy donor procured in an elective setting, and optimal transplant circumstances – coordinated donor–recipient (D–R) operations with minimal cold preservation. More recent data suggest that some of the incremental benefit of LDKT may derive from simply shortening and/or pre-empting dialysis (4,5).

While excellent organ quality and ideal transplant conditions eliminate many of the known factors that compromise early graft function (EGF), poor EGF still occurs after LDKT, albeit less frequently than after DDKT. It is unclear whether the well-known risk factors for DGF after DDKT are still relevant in the LDKT setting. Furthermore, although there is voluminous literature discussing the impact of early function, particularly DGF, on outcomes after DDKT, much less has been written regarding the impact of early function in the setting of LDKT.

We undertook this retrospective study of 469 LDKT performed at the University of California San Francisco between January 1, 1997 and December 31, 2001 to determine the incidence of and risk factors for poor EGF and to understand its impact on transplant outcomes. We have defined poor EGF as the occurrence of either delayed graft function (DGF), defined by a requirement for dialysis within 1 week of transplantation, or slow graft function (SGF), defined by serum creatinine (Cr) ≥ 3.0 mg/dL 5 days after transplantation without dialysis.

Materials and Methods

Patients and data collection

Retrospective medical record review was performed to collect donor, recipient, transplant and post-transplant data for 470 LDKTs performed at the University of California San Francisco between January 1, 1997 and December 31, 2001. One transplant was excluded owing to vascular thrombosis within 24 h of transplantation.

Donor evaluation

Individuals interested in kidney donation were initially screened by ABO typing and a detailed medical, family, and social history questionnaire. Donor HLA typing and T-cell cytotoxic and flow cytometric cross-matching were performed to assess donor–recipient (D–R) compatibility. Medical and laboratory evaluation then proceeded in parallel to assess medical suitability and adequacy of renal function based upon serum blood urea nitrogen, serum creatinine (Cr), and 24-h urine collection. Psychosocial suitability was assessed by interview with a kidney transplant social worker. Intravenous pyelography and renal arteriography to define anatomy represented the final phase of donor evaluation.

Kidney procurement and transplantation procedures

Donor and recipient operations were performed simultaneously in adjoining operating rooms. Kidneys were procured by standard open technique using a flank incision until November 1999 when laparoscopic donor nephrectomy was first performed at our institution (6). In 2000 and 2001, the laparoscopic technique accounted for 37% and 62%, respectively, of LDKT procurements. Immediately after procurement, kidneys were placed on ice, brought into the recipient room, flushed with lactated Ringer's solution containing heparin, sodium bicarbonate, and procainamide and then transplanted.

Kidneys were typically placed in the recipient's iliac fossa through an extraperitoneal approach with vascular anastomoses to the external iliac vessels. An intra-abdominal approach was used for three recipients, all undergoing their third kidney transplant. Extravesicular ureteroneocystostomy was performed, usually without a stent. Intravenous mannitol and furosemide were infused before graft reperfusion. Warm ischemia time is equivalent to anastomotic time.


Essentially all patients received steroids (intravenous methyl-prednisolone first administered in the operating room followed by oral prednisone) and mycophenolate mofetil (typically 2 g orally per day adjusted for abnormal hematologic parameters and/or clinical symptoms). Nearly all recipients (451 of 469; 96%) also received calcineurin inhibitors (CNI), either cyclosporine (216 of 451; 48%) or tacrolimus (235 of 451; 52%), typically initiated in the presence of good renal function. Cyclosporine was dosed to achieve 12-h trough levels of 200–300 µg/L; tacrolimus was dosed to achieve 12-h trough levels of 8–15 µg/L for the first three post-transplant months. During the 5-year period, an increasing percentage of recipients were initiated on tacrolimus: 31% in 1997 to 65% in 2001. Of the 18 recipients who were not initiated on any CNI, 16 were participants in various CNI-sparing studies and two lost nonfunctioning grafts, one to transplant nephrectomy on post-transplant day 9 and the other to recipient death on post-transplant day 12. Anti-CD25 antibody therapy was given to 25% of the patients (n = 119) and a minority of patients (n = 25; 5.3%) received a depleting antibody preparation (ATGAM, OKT3, or rabbit thymoglobulin) as induction immunosuppression secondary to high immunologic risk (high-panel reactive antibody and/or loss of previous transplant secondary to rejection).

Diagnosis of rejection

All rejection episodes were biopsy-proven, identified by review of biopsy reports within 1 year of LDKT. Typically, clinical and/or laboratory evidence of graft dysfunction prompted ultrasound and Doppler examination of the allograft and biopsy was immediately performed in the absence of technical or mechanical complication. During DGF, ‘protocol’ biopsies were performed at 5–10-day intervals. Of the 22 recipients with DGF, only three did not undergo a single biopsy because their duration of DGF was short (last dialysis days were 1, 3, and 5 days after transplantation). In addition, recipients who enrolled in studies underwent protocol biopsy as dictated by the relevant study protocol.

Statistical analysis

The EGF groups were compared for demographic and other characteristics using Kruskal–Wallis or Chi-squared tests for continuous and noncontinuous variables, respectively. The odds ratios (ORs) for poor EGF associated with donor, recipient, and transplant characteristics were determined by logistic regression analysis. The hazard ratios (HRs) for rejection associated with donor, recipient, and transplant characteristics were determined by Cox proportional hazards analysis. Variables in univariate models which were either significantly associated (p < 0.05) or which exhibited a trend towards association (p < 0.10) with poor EGF or rejection were selected for multivariate analysis. For both the logistic regression and Cox proportional hazards models, the linearity of continuous variables was examined using quartiles and by testing for quadratic effects. No evidence of nonlinearity was found. Proportional hazards assumptions were tested by including interactions of each predictor with log of time. To satisfy the proportional hazards assumption in multivariate models, we stratified the analysis on nonproportional variables. Patient survival, graft survival (with or without death censored), and time to first rejection within 1 year of LDKT were determined by Kaplan-Meier analysis and compared using log-rank and Wilcoxon tests. All analyses were performed using SAS, version 8.2 (SAS Institute, Cary, NC).


LDKT donor, recipient, and transplant characteristics

Between January 1, 1997 and December 31, 2001, the University of California San Francisco Kidney Transplant Service performed 470 living donor kidney transplants. One graft thrombosed within 24 h of transplantation and was therefore excluded from this retrospective study. Of the 469 remaining transplants, 428 (91%) occurred in adult recipients (≥18 years) and 41 occurred in pediatric recipients (<18 years of age). Thirty-three recipients (7%) were undergoing their second transplant while four recipients (1%) were undergoing their third kidney transplant. There were 346 (74%) related and 123 (26%) unrelated donor–recipient (D–R) pairs. The most common related donor was a sibling, while the most common unrelated donor was a spouse. There were 53 (11%) haploidentical transplants. Nearly 22% of kidneys (102 of 469) were procured laparoscopically.

Our cohort of 469 transplants was divided into three groups by EGF. Twenty-two recipients (4.7%) experienced DGF, defined as requiring dialysis during the first week after transplantation; 50 recipients (10.7%) experienced SGF defined as serum Cr ≥ 3.0 mg/dL on post-transplant day five without any dialysis (7); and the remaining 397 recipients (85%) experienced immediate graft function (IGF). Comparison of the three EGF groups for donor, recipient, and transplant characteristics showed differences in only four characteristics: recipient body mass index (BMI) (p = 0.029), recipient etiology of renal disease (p = 0.015), unrelated D–R pair (p = 0.004), and warm ischemia time (WIT) (p = 0.013) (Table 1). There was a trend towards older donors (p = 0.071) in the poor EGF groups. The three groups did not differ in donor gender, ethnicity, or mode of kidney procurement; recipient age, gender, ethnicity, dialysis status, sensitization, or history of previous transplantation; and transplant year or degree of HLA mismatch (MM).

Table 1.  Donor, recipient, and transplant characteristics by EGF groups: UCSF Living Donor Kidney Transplants 1997–2001
CharacteristicIGF (n = 397)SGF (n = 50)DGF (n = 22)p-value*
Donor factors    
 Age (years)
  Mean40.8 ± 10.743.5 ± 9.344.1 ± 11.20.071
  >50 year78 (20%)13 (26%)7 (32%)0.25
 Gender 0.88
  Female244 (61%)29 (58%)13 (59%) 
  Male153 (39%)21 (42%)9 (41%) 
 BMI (kg/m2)25.6 ± 4.126.1 ± 3.825.3 ± 3.20.57
 Ethnicity 0.88
  African-American33 (8%)6 (12%)1 (5%) 
  Asian47 (12%)4 (8%)2 (9%) 
  Caucasian251 (63%)33 (66%)17 (77%) 
  Hispanic63 (16%)7 (14%)2 (9%) 
  Other3 (1%)0 (0%)0 (0%) 
 Kidney procurement 0.23
  Laparoscopic84 (21%)15 (30%)3 (14%) 
  Open313 (79%)35 (70%)19 (86%) 
Recipient factors
 Age (years)
  Mean40.8 ± 15.943.2 ± 13.547.0 ± 18.10.19
  <1838 (10%)2 (4%)1 (5%)0.29
  18–50235 (59%)32 (64%)11 (50%) 
  >50124 (31%)16 (32%)10 (45%) 
 Gender 0.73
  Male (n = 284)231 (58%)32 (64%)11 (50%) 
  Female (n = 185)166 (42%)18 (36%)11 (50%) 
 BMI (kg/m2)24.8 ± 5.126.2 ± 5.126.6 ± 4.20.029
 Ethnicity 0.51
  African-American31 (8%)6 (12%)2 (9%) 
  Asian43 (11%)4 (8%)2 (9%) 
  Caucasian227 (57%)32 (64%)11 (50%) 
  Hispanic74 (19%)5 (10%)7 (32%) 
  Other22 (6%)3 (6%)0 (0%) 
 Disease etiology 0.015
  Diabetes64 (16%)13 (26%)10 (45%) 
  Hypertension/glomerulosclerosis57 (14%)4 (8%)1 (5%) 
  Glomerulonephritis117 (29%)18 (26%)3 (14%) 
  Polycystic40 (10%)4 (8%)0 (0%) 
  IgA nephropathy30 (8%)7 (14%)3 (14%) 
  Congenital/urological35 (9%)2 (4%)1 (5%) 
  Unknown/other54 (14%)2 (4%)4 (18%) 
 Disease etiology 0.0009
  Diabetes64 (16%)13 (26%)10 (45%) 
  Non-diabetes333 (84%)37 (74%)12 (55%) 
 Dialysis mode 0.40
  None79 (20%)6 (12%)3 (14%) 
  Hemodialysis245 (62%)35 (70%)17 (77%) 
  CAPD73 (18%)9 (18%)2 (9%) 
 Peak PRA > 30%1 0.27
  Yes12 (5%)13 (10%)12 (12%)1 
  No239 (95%)127 (90%)115 (88%)1 
 Previous kidney transplant 0.54
  Yes31 (8%)3 (6%)3 (14%) 
  No366 (92%)47 (94%)19 (86%) 
Transplant factors
 Year of Transplant 0.13
  199777 (19%)7 (14%)6 (27%) 
  199877 (19%)9 (18%)2 (9%) 
  199972 (18%)5 (10%)6 (27%) 
  200088 (22%)12 (24%)4 (18%) 
  200183 (21%)17 (34%)4 (18%) 
CharacteristicIGF (n = 397)SGF (n = 50)DGF (n = 22)p-value*
  1. *Kruskal–Wallis for continuous variables; Chi-squared for categorical variables.

  2. 1Peak PRA was unavailable for 171 transplants (36%): 146 in IGF (n = 251), 20 in SDF (n = 30), and five in DGF (n = 17) groups.

  3. 2HLA MM was unavailable for 30 transplants (6%): 25 in IGF (n = 372) and five in DGF (n = 17) groups.

  4. 3WIT data was unavailable for 28 transplants (6%): 24 in IGF (n = 373), two in SGF (n = 48), and two in DGF (n = 20) groups.

Transplant factors (continued)
 D–R BMI ratio (mean ± SD)1.08 ± 0.271.03 ± 0.210.97 ± 0.230.12
 HLA MM2 0.18
  047 (13%)25 (10%)21 (6%)2 
  1–3212 (57%)223 (46%)210 (59%)2 
  >3113 (30%)222 (44%)26 (35%)2 
 D–R pair relationship 0.004
  Related302 (76%)34 (68%)10 (45%) 
  Unrelated95 (24%)16 (32%)12 (55%) 
 Warm ischemia time (min)3
  Mean ± SD30.9 ± 8.235.3 ± 12.033.7 ± 7.00.013
  >30 min177 (47%)331 (65%)311 (55%)30.073

Risk factors for poor EGF after LDKT

Logistic regression analysis was used to determine the ORs associated with donor, recipient, and transplant characteristics for poor EGF: the occurrence of either SGF or DGF rather than IGF (Table 2). Univariate models showed that donor age [OR 1.03 per year; 95% confidence interval (CI) = 1.00–1.05; p = 0.037], recipient BMI (OR 1.06 per 1.0 increment; 95% CI = 1.01–1.11; p = 0.017), diabetic etiology of renal disease (OR 2.44; 95% CI = 1.39–4.29; p = 0.0019), unrelated D–R pair (OR = 2.02; 95% CI = 1.19–3.43; p = 0.0088), and WIT, considered as a continuous variable (OR = 1.05 per min increment; 95% CI = 1.02–1.07; p = 0.0014) or stratified at >30 min (OR = 1.79; 95% CI = 1.05–3.04; p = 0.031) were significantly associated with poor EGF (Table 2 A). Older recipient age (OR 1.01; p = 0.080), higher D–R BMI ratio (OR 0.36; p = 0.069), and >3 HLA MM (OR = 1.65; p = 0.067) exhibited trends (p < 0.10) towards poor EGF. No other donor, recipient, or transplant factors were positively or negatively associated with poor EGF.

Table 2.  Relative risk of donor, recipient, and transplant characteristics for the development of SGF or DGF vs. IGF: UCSF Living Donor Kidney Transplants 1997–2001
CharacteristicOdds ratio95% CIp-value
  1. CI = confidence interval.

A. Univariate models
 Donor age (per year)1.031.00–1.050.037
 Donor age >50 years1.570.89–2.790.12
 Donor male gender1.140.68–1.900.62
 Donor ethnicity: AA vs. all others1.190.50–2.800.69
 Donor BMI (per 1.0 increment)1.020.96–1.080.60
 Kidney procurement: laparoscopic vs. open1.240.69–2.230.47
 Recipient age (per year)1.011.00–1.030.080
 Recipient age ≥ 50 years1.100.64–1.910.72
 Recipient male gender1.200.71–2.010.49
 Recipient BMI (per 1.0 increment)1.061.01–1.110.017
 D–R BMI ratio0.360.12–1.080.069
 Recipient ethnicity: AA vs. all others1.110.67–1.850.69
 Recipient disease: diabetes vs. all others2.441.39–4.290.002
 Recipient dialysis: yes vs. no1.740.83–3.650.14
 Recipient PRA > 30%2.370.79–7.080.12
 Recipient with previous transplant1.070.43–2.670.88
 Transplant year
  1998 vs. 19970.850.36–2.010.70
  1999 vs. 19970.900.38–2.150.82
  2000 vs. 19971.080.49–2.380.85
  2001 vs. 19971.500.70–3.200.30
 >3 HLA MM1.650.97–2.810.067
 Unrelated D–R pair2.021.19–3.430.009
 WIT (per min)1.051.02–1.070.001
 WIT > 30 min1.791.05–3.040.031
B. Multivariate model
 Donor age1.031.00–1.050.079
 Recipient age0.990.97–1.010.42
 Recipient BMI1.030.98–1.090.24
 Recipient diabetes2.221.13–4.350.021
 >3 HLA MM1.350.64–2.850.43
 Unrelated D–R pair1.690.77–3.740.19

A multivariate regression model was then constructed including all characteristics associated with poor EGF (p < 0.10) (Table 2B). Recipient BMI and WIT as continuous variables were preferentially selected over D–R BMI ratio and WIT >30 min, respectively. Diabetic etiology of renal disease and increased WIT retained their significant explanatory power for poor EGF in the multivariate model.

Patient and graft survival after LDKT according to EGF quality

Overall, patient survival for our cohort of 469 LDKT recipients with a mean follow up of 31 months was 96.6%. There were 16 deaths in total: 12 in the IGF group and two each in the SGF and DGF groups. Interestingly, all four deaths in the SGF and DGF groups occurred early,– within 6 months following transplantation, compared with three of the 12 deaths in the IGF group. The causes of early death were cerebrovascular accident and sepsis for the DGF patients, sepsis and cardiac arrest for the SGF patients, and sepsis, cardiac arrest, and delayed hemorrhage after kidney biopsy for the IGF patients. Kaplan-Meier analysis of patient survival by EGF showed significant differences in 1-year patient survival among the three early function groups (p = 0.029), but no significant differences in overall survival (p = 0.15) (Figure 1).

Figure 1.

Kaplan-Meier analysis of patient survival by EGF. The EGF groups differ in 1-year patient survival (p-value = 0.029) but not in overall patient survival with a mean follow up of 30 months (p-value = 0.15).

Graft survival for our cohort of living donor kidney transplant was 91.9% during the study period. There were 38 graft failures: five, six, and 27 in the DGF, SGF, and IGF groups, respectively. Two transplants, both in the DGF group, never functioned. One graft was lost when the recipient died of a cerebrovascular accident 12 days after transplantation; the second graft was removed 9 days after transplantation after ultrasound and renal perfusion scintigraphy examinations suggested absence of cortical perfusion, which was confirmed by biopsy. Of the 36 grafts that functioned and subsequently failed, 13 were lost within 1 year of transplantation (one, three, and nine in the DGF, SGF, and IGF groups, respectively). While eight of 13 grafts lost within 1 year were secondary to recipient death, only one of the 23 grafts lost after 1 year was secondary to recipient death.

Kaplan-Meier analysis of graft survival demonstrated a trend towards worse 1-year graft survival (p = 0.065) and a significant decrement in overall graft survival (p = 0.044) for the poor EGF groups (data not shown). However, analysis of graft survival with death censored showed no differences among the EGF groups either at 1 year (p = 0.26) or overall (p = 0.39) (Figure 2). Finally, analysis including only grafts functioning at 1 year showed no differences among the EGF groups in graft survival with (p = 0.79) or without death censored (p = 0.73; data not shown).

Figure 2.

Kaplan-Meier analysis of death-censored graft survival by EGF. The EGF groups did not differ significantly in either 1-year (p-value = 0.26) or overall death-censored graft survival (p-value = 0.39).

One-year LDKT graft function according to EGF quality

Three parameters were used to assess graft function 1 year after transplantation: serum creatinine (Cr), Cr clearance, and delta (Δ) Cr (1 year Cr minus six month Cr) (Table 3). This analysis excluded 40 of 469 grafts (8.5%): two grafts which never functioned, 13 grafts lost during the first year, and 25 grafts with missing data. There were no apparent differences among the DGF, SGF, and IGF groups with respect to 1-year Cr (p = 0.21), Cr clearance (p = 0.94), or ΔCr (p = 0.30).

Table 3.  One year graft function by EGF: UCSF Living Donor Kidney Transplants 1997–2001
Graft function
(n = 375)
(n = 46)
(n = 19)

  1. *Kruskal–Wallis.

1-year Cr 1.4 ± 0.6 1.7 ± 0.9 1.4 ± 0.50.21
1-year Cr66.6 ± 21.567.2 ± 26.271.8 ± 30.50.94
Delta-Cr0.01 ± 0.6 0.2 ± 0.7−0.1 ± 0.30.30

Incidence of rejection within 1 year of LDKT according to EGF quality

In our cohort of 469 living donor kidney transplants, 95 recipients (20.3%) had at least one episode of biopsy-proven rejection within 1 year of transplantation. Ten recipients – one, three, and six in the DGF, SGF, and IGF groups, respectively – experienced more than one episode of rejection. The incidence of rejection at 1 year was significantly less for the IGF group (16%) compared with both the SGF (40%) and DGF (50%) groups (p < 0.0001 for both comparisons). The incidence of rejection between the SGF and DGF groups was, however, comparable (p = 0.32) (Figure 3). As the majority of recipients (50 of 91; 55%) who rejected did so during the first month after transplantation, the time to rejection curves diverge early and remain essentially parallel throughout the study period.

Figure 3.

Kaplan-Meier analysis of time to rejection within 1 year of transplantation by EGF. Recipients with either SGF and DGF experienced substantially more rejection compared with recipients with IGF (p-value < 0.0001 for both comparisons). The DGF and SGF groups however, did not differ (p-value = 0.32).

Risk factors for rejection within 1 year of LDKT

Cox proportional hazards analysis identified donor, recipient, transplant, and post-transplant risk factors for rejection within 1 year of transplantation (Table 4A). Donor age, a characteristic not significantly associated with poor EGF, strongly predisposed to rejection both as a continuous variable (HR = 1.04 per year increment; 95% CI = 1.02–1.06; p < 0.0001) or stratified as >50 years (HR = 2.55; 95% CI = 1.68–3.87; p < 0.0001). Transplants with >3 HLA MMs (OR = 2.16; 95% CI = 1.41–3.29; p = 0.0004) or those involving an unrelated D–R pair (OR = 1.95; 95% CI = 1.29–2.96; p = 0.0016) were approximately twice as likely to experience rejection. As these two variables, unrelated D–R pair and >3 HLA MM, did not fulfil proportional hazards assumptions, their relative risks for rejection were calculated and found to be highly significant by Fisher's exact test (>3 HLA MM: RR = 1.95, p = 0.001; unrelated D–R pair: RR = 1.75, p = 0.005). The most robust risk factors for rejection after LDKT were, however, poor EGF: DGF (HR = 4.84; 95% CI = 2.55–9.19; p < 0.0001) and SGF (HR = 3.08; 95% CI = 1.86–5.09; p < 0.0001). Interestingly, none of the significant risk factors for poor EGF – recipient BMI, diabetic etiology of renal disease, and WIT – were associated with rejection. A multivariate Cox model including all variables associated with rejection (p < 0.10) and stratified on D–R relationship demonstrated that four variables – donor age >50 years, >3 HLA MM, DGF, and SGF – retain significant explanatory power for rejection (Table 4B).

Table 4.  Cox models of time to rejection within 1 year of transplantation: UCSF Living Donor Kidney Transplants 1997–2001
CharacteristicHazard ratio95% CIp-value
  1. CI = confidence interval.

  2. 1Variable did not fulfill proportional hazards assumption.

A. Univariate models   
 Donor age1.041.02–1.06<0.0001
 Donor age >50 years2.551.68–3.87<0.0001
 Donor BMI (per 1.0 increment)1.030.97–1.08 0.37
 Kidney procurement: laparoscopic vs. open0.780.46–1.31 0.43
 Recipient age1.000.99–1.02 0.64
 Recipient age >50 years1.180.77–1.80 0.45
 Recipient gender1.170.77–1.77 0.45
 Recipient ethnicity: AA vs. all others1.000.48–2.05 0.99
 Recipient BMI (per 1.0 increment)1.030.99–1.08 0.16
 Recipient disease: diabetes vs. all others11.260.77–2.07 0.36
 Recipient PRA >30%0.800.25–2.53 0.69
 Recipient with previous transplant0.890.41–1.92 0.76
 Year of transplant
  1998 vs. 19970.790.40–1.54 0.48
  1999 vs. 19971.080.58–2.03 0.81
  2000 vs. 19970.880.48–1.64 0.70
  2001 vs. 19970.950.52–1.76 0.88
 D–R BMI ratio0.720.29–1.78 0.48
 >3 HLA MM12.161.41–3.29 0.0004
 Unrelated D–R pair11.951.29–2.96 0.0016
 WIT (per min)1.000.98–1.02 0.96
 WIT >30 min1.210.80–1.84 0.36
 DGF vs. IGF4.842.55–9.19<0.0001
 SGF vs. IGF3.081.86–5.09<0.0001
 CYA vs. FK10.920.60–1.42 0.72
B. Multivariate model
 DGF vs. IGF4.271.98–9.22 0.0002
 SGF vs. IGF2.961.77–4.95<0.0001
 Donor age >50 years2.481.59–3.88<0.0001
 >3 HLA MM2.351.39–3.96 0.0015

Patient survival, graft survival, and 1-year graft function according to rejection status

Finally, Kaplan-Meier analyses for patient and graft survival were repeated according to rejection rather than EGF status. Rejectors and nonrejectors did not exhibit any differences in patient survival during the follow-up period (p = 0.80; data not shown). However, rejection during the first year after transplantation was associated with a significant decrement in overall graft survival (data not shown) and death-censored graft survival (Figure 4). This difference in graft survival was corroborated by analysis of 1-year graft function by rejection status (Table 5). Recipients who did not experience rejection exhibited consistently superior graft function as assessed by 1-year Cr (1.4 vs. 1.8 mg/dL; p < 0.0001), 1-year Cr clearance (68.3 vs. 61.0 mL/min; p = 0.011), and Δ Cr (0.0 vs. 0.2 mg/dL; p = 0.018).

Figure 4.

Kaplan-Meier analysis of death-censored graft survival by rejection status. Recipients who experienced rejection during the first year after transplantation had lower functional graft survival compared with those who did not reject (p-value = 0.0007).

Table 5.  One year graft function by rejection status: UCSF Living Donor Kidney Transplants 1997–2001

Graft function parameter
No rejection
(n = 374)
(n = 95)

  1. *Mann–Whitney.

1-year Cr 1.4 ± 0.6 1.8 ± 0.8<0.0001
1-year Cr Cl68.3 ± 22.061.0 ± 23.7 0.011
Delta-Cr 0.0 ± 0.5 0.2 ± 0.6 0.018


The new millennium, the year 2001, signaled a new era in kidney transplantation where the number of transplants from living donors exceeded that from deceased donors. It is well-recognized that LDKT results in superior EGF and overall transplant outcomes than DDKT (1–3,8,9). While much is known regarding the incidence and impact of poor EGF (most commonly DGF but also SGF) in the setting of DDKT (7,10–14), little has been written on these issues in the setting of LDKT. The living-donor graft provides a unique opportunity to investigate why poor EGF may develop in spite of optimized donor and transplant characteristics and its impact on important post-transplant outcomes.

In our retrospective study with a large, modern 5-year cohort of 469 LDKTs, the incidence of DGF and SGF were 4.7% and 10.7%, respectively. As expected, these incidences are substantially lower than the reported rates of 20–29% for DGF (as reviewed by Halloran and Hunsicker, 2001 (10)] and 27–31% for SGF after DDKT (7,15). In the absence of risk factors which dominate the DDKT setting, such as donor medical comorbidities, brain death, compromised renal function, and cold ischemia time, only two factors emerged as independently and significantly associated with poor EGF in the LDKT setting: diabetic etiology of renal disease and WIT. Increased donor age, widely considered to be the most potent risk factor for DGF and SGF after DDKT (10–13,16,17), did not achieve significance in our cohort of medically healthy live donors. Not only the presence, but also the potency, of diabetic etiology as a risk factor for poor EGF after LDKT is quite intriguing, as it has not been identified as such after DDKT (10,13). The multivariate analyses that we have performed do not support several possible hypotheses to explain this association: that diabetic recipients have atherosclerotic disease predisposing to longer anastomotic times, that diabetic recipients have higher BMIs, and that diabetic recipients are more likely to be older, have older donors or have unrelated donors. The explanation(s) more likely reflects physiologic rather than technical or epidemiologic differences between diabetic and nondiabetic recipients and certainly deserves additional study.

Next, we determined the impact of poor EGF on important outcomes for our LDKT cohort. While poor EGF groups demonstrated worse 1-year patient survival, overall patient survival did not differ among the three groups. These results are consistent with the association, in the DDKT setting, between DGF and an increased probability of early death (<6 months after transplantation) (18). In contrast, functional (death-censored) graft survival, both at 1 year and overall, however, did not differ among the three EGF cohorts. These results regarding the impact of EGF on functional outcomes after LDKT differ significantly from those previously reported in association with DDKT. A substantial body of literature indicates that DGF after DDKT impairs long-term graft outcomes (11–13). Furthermore, Humar, in his study of DGF, SGF, and IGF, reported that the durability of DDKTs exactly paralleled the quality of EGF (7,15). The IGF group with optimal EGF had the best functional graft survival, the intermediate SGF group had intermediate survival, and the DGF group with the poorest EGF had the worst graft survival. Others have similarly suggested that poor EGF after LDKT, in contrast to DDKT, does not impair long-term graft survival (19,20). One could hypothesize that the severity of injury required to produce poor EGF after LDKT compared with DDKT would be substantially greater, which might translate into worse functional graft survival. Our data however, suggest that the optimal kidney quality and transplant circumstances of LDKT may ameliorate the deleterious impact of poor EGF on functional graft survival.

Although poor EGF did not appear to impair functional graft survival for our LDKT cohort, DGF and SGF were the most potent risk factors for AR within 1 year of transplantation. Our results essentially mirror those previously reported by Humar in his DDKT cohort of IGF, SGF, and DGF, where the incidence of AR exactly paralleled EGF quality (7,15): recipients with DGF experienced the most; those with IGF experienced the least; and those with SGF were intermediate. The relationship between DGF and AR after DDKT has been both well established and well accepted (10,12,13,21). Interestingly, the incidence of AR with DGF after both LDKT and DDKT was entirely comparable (7,10,12). We postulate that the occurrence of DGF signals a certain degree of allograft injury. Injury evokes inflammation, which enhances immune recognition and, in turn, predisposes to rejection. If sufficient injury to produce DGF or even SGF occurs, the initial superior graft quality characteristic of LDKTs cannot reduce the subsequent risk of AR.

In our LDKT cohort, interestingly, donor age >50 years also emerged as an independent and powerful risk factor for rejection. One could hypothesize that, in general, kidneys from older donors may tolerate procurement and transplantation less well, sustaining greater ‘injury’ than kidneys from younger donors, thereby predisposing older kidneys to rejection. The final independent risk factor for AR after LDKT which emerged was, not surprisingly, D–R immunologic similarity. Recipient BMI, diabetic disease etiology, and increased WIT, factors predictive of poor EGF, were not directly associated with rejection. Therefore, although poor EGF substantially predisposed to AR, risk factors for poor EGF were not risk factors for AR.

Lastly, we assessed the impact of AR within 1 year of transplantation on death-censored graft survival and 1-year graft function. All of our analyses yielded consistent results indicating significantly worse outcomes for rejectors compared with nonrejectors. Rejection has been identified as an independent risk factor for graft failure after DDKT (10,12,13,17,19,21) and LDKT (19,20,22). Indeed, Matas and colleagues have reported that, for their cohort of LDKTs transplanted between 1985 and 1996, AR was the only significant risk factor for late graft failure (19). Our analysis of a LDKT cohort undergoing transplantation in a more compressed, 5-year time period confirmed that the occurrence of AR remains the prime determinant of long-term functional graft survival even in the current era of ‘modern’ immunosuppression. Therefore, poor EGF after LDKT is clearly an indirect but substantial threat to functional graft survival. Our data provides compelling motivation to consider strategies to prevent poor EGF or, perhaps more realistically, to intensify immunosuppression for LDKTs involving donors >50 years of age, >3 HLA MMs, and/or LDKTs which experience poor EGF. Prevention of acute rejection should maximize the longevity of the precious living donor kidney resource. In this era of hierarchical choices for immunosuppression, transplant centers and physicians, armed with data and knowledge regarding the risks posed by various donor, recipient, transplant, and post-transplant characteristics, can better select, within the context of their general immunosuppression philosophy, the optimal immunosuppression strategy for specific LDKT donor and recipient profiles.


Funding sources: TVB is supported by a grant from the American College of Surgeons. We thank TeleResults® for assisting us with patient data collection.