Recent therapeutic strategies in renal transplantation includes attempts to reduce/avoid calcineurin inhibitor nephrotoxicity, reduce/avoid corticosteroids, and maximize the use of adjunctive maintenance antiproliferative agents (inosine monophosphate dehydrogenase [IMPDH] or target of rapamycin [TOR] inhibitors), with the respective goals of preventing/reducing long-term (vascular) chronic allograft injury (CAI), detrimental effects of continuous corticosteroid use, and incidence/severity of acute rejection (AR) while simultaneously achieving more favorable long-term patient and graft survival (1–5). A prospective, open-labeled randomized clinical trial of 150 adult, primary kidney transplant recipients was performed at our center comparing mycophenolate mofetil (MMF) versus enteric-coated mycophenolate sodium (EC-MPS). The plan to compare these two IMPDH inhibitors was based on possible avoidance of gastrointestinal (GI) toxicity using EC-MPS versus MMF in a reduced maintenance tacrolimus dose (rTd)-based regimen, which might further allow higher dosing of the former agents (and, thus, achieve greater therapeutic efficacy), especially as tacrolimus has its own inherent GI side effects. This study was the first randomized trial to compare these two IMPDH inhibitors using a reduced dose tacrolimus-based regimen and steroid elimination (5), with the results at 1 year posttransplant showing low biopsy-proven acute rejection (BPAR), acceptably high renal function, and no differences in the incidence of symptomatic GI side effects between the two groups. Results of this trial at 4 years posttransplant are reported in this study, examining whether rTd used with maximized long-term maintenance of IMPDH inhibitors provides efficacious and safe posttransplant immunosuppression in the planned absence of corticosteroids.
Patient Demographics and Early Outcome
Distribution of the baseline demographics and early outcome variables are listed for each study group in Table 1. There were no significant differences between the two groups regarding any of the baseline characteristics. Although a slightly higher (nonsignificant) percentage of patients in the EC-MPS arm received nonstandard deceased donor (DD) kidneys (i.e., donation after cardiac death, pediatric en bloc, adult double kidneys, or a single, expanded criteria donor kidney), a slightly lower (nonsignificant) percentage of patients in the EC-MPS arm received a living donor kidney. Twenty percent (30/150) of patients in this study received nonstandard DD kidneys.
The incidence of delayed graft function (DGF) was low: 1.3% (1/75) vs. 4.0% (3/75) in the MMF and EC-MPS arms, respectively (nonsignificant). One additional patient in the MMF arm had a never functioning (ruptured) kidney graft. Three patients having DGF (one in group A and two in group B) received a donation after cardiac death kidney; one additional DGF patient (EC-MPS arm) received adult double kidneys.
Immunosuppressive Doses and Trough Levels
All patients received the scheduled two courses of daclizumab (Dac) and three courses of antithymocyte globulin (ATG) as induction therapy. The four patients having DGF received 3 to 11 additional courses of ATG as described previously (5).
Mean doses of rTd, MMF, and EC-MPS, serum trough concentrations, white blood cell count, and percentages of patients receiving 100% and ≥75% of planned dosages of MMF and EC-MPS are shown for the two study groups in Table 2.
Mean tacrolimus trough levels seemed to have reached those planned, with mean levels at month 48 being 5.96±0.27 (n=55) in group A vs. 5.92±0.22 (n=53) in group B (not significant). Of note, the mean tacrolimus trough level was significantly higher at month 24 in the EC-MPS group (P=0.04). Among 55 patients in group A receiving tacrolimus at month 48, the frequency distribution of tacrolimus levels was <4 (n=6), ≥4 and <6 (n=24), and ≥6 (n=25). Among 53 patients in group B receiving tacrolimus at month 48, the frequency distribution of tacrolimus levels was <4 (n=4), ≥4 and <6 (n=26), and ≥6 (n=23).
In general, there was equivalence between the groups in meeting maximized IMPDH inhibitor doses; although the percentage receiving a full dose was somewhat higher in group B (with statistical significance at month 24), there were no significant differences in percentage receiving ≥75% of the planned dose. Mean dosage of MMF and EC-MPS at month 48 posttransplantation (Table 2) was 1207±68 (n=52) and 943±46 (n=58), respectively. Among the 52 patients receiving MMF (group A) at month 48, the frequency distribution of doses was <1500 (n=31), ≥1500 but <2000 (n=12), and 2000 (n=9). Among the 58 patients receiving EC-MPS (group B) at month 48, the frequency distribution of doses was <1080 (n=33), ≥1080 but <1440 (n=9), and 1440 (n=16). Thus, the percentage taking ≥75% of the prescribed dose at month 48 was 40.4% (21/52) in group A versus 43.9% (25/57) in group B.
Percentages having MMF or EC-MPS withheld are shown in Table 3. In group A, reasons for withholding of MMF included leukopenia (n=9), insurance issues (n=5), GI symptoms (n=3), wound infection (n=1), and unknown (n=1). In group B, reasons for withholding of EC-MPS included leukopenia (n=8), AR (n=2, switched to MMF), and GI symptoms (n=1). Time to withholding of MMF and EC-MPS for all 17 patients with leukopenia occurred during the first 12 months posttransplant. Conversely, time-to-withholding of MMF for the five patients having insurance issues occurred at or beyond 36 months posttransplant. Reasons for discontinuing MMF in group A included leukopenia (n=1), insurance issues (n=4), and GI symptoms (n=3). Reasons for discontinuing EC-MPS in group B included malignancy (n=1), AR (n=3, switched to MMF), and GI symptoms (n=2).
Group A had higher percentages of patients receiving steroids at 24 and 36 months (P=0.04 and 0.03, respectively). Note that only 13 of 60 (22%) patients in group A and 8 of 60 (13%) patients in group B (18% overall) remained on corticosteroids at 48 months postoperatively (P=0.23). The percentage remaining off steroids throughout 48 months of posttransplant follow-up was 68% (51/75) and 75% (56/75) in groups A and B, respectively (P=0.37). Reasons for continuing/restarting corticosteroids included DGF or slow graft function (five in group A and six in group B), focal segmental glomerulosclerosis as original diagnosis (five in group A and three in group B), occurrence of BPAR (12 in group A and 8 in group B), and other (two in group A and two in group B). There were no observed differences in white blood cell counts between groups.
There was no difference between groups in the rate of developing a first BPAR episode during the first 48 months posttransplant: 14 of 75 (19%) in group A versus 13 of 75 (18%) in group B including borderline (P=0.86), and 8 of 75 (11%) in group A versus 10 of 75 (14%) in group B excluding borderline (P=0.60) (Table 4). Overall, 27 patients (18%) developed a first BPAR during the first 48 months posttransplant; excluding borderline cases, only 18 patients (12%) developed a first rejection episode.
Among 14 group A patients experiencing BPAR during the first 48 months posttransplant, 5 (35.7%) were white, 5 (35.7%) were Hispanic, and 4 (28.6%) were African American. Among 13 group B patients experiencing BPAR during the first 48 months posttransplant, two (15.4%) were white, five (38.5%) were Hispanic, and six (46.1%) were African American. Overall, 74% (20/27) developing BPAR were minorities.
Three patients in group A and one patient in group B experienced a second BPAR episode (grades were IA, IA with C4d+, and IA with donor-specific antibodies in group A, and IB in group B). All second BPAR episodes were treated with antilymphocyte therapy; the patient with donor-specific antibodies also received plasmapheresis and rituximab. Overall, only six patients (four in group A and two in group B) developed a C4d+ rejection during 48 months posttransplant.
Among 14 group A patients experiencing BPAR during the first 48 months posttransplant, the frequency distribution of MMF doses at or within 3 months of BPAR was <1500 (n=11), ≥1500 but <2000 (n=3), and 2000 (n=0). Among 13 group B patients experiencing BPAR during the first 48 months posttransplant, the frequency distribution of EC-MPS doses at or within 3 months of BPAR was <1080 (n=6), ≥1080 but <1440 (n=2), and 1440 (n=5).
Fourteen of 27 patients (5/14 in group A and 9/13 in group B) received antilymphocyte therapy in treating the first BPAR; two thirds (18/27) remained on corticosteroids at 48 months (or immediately preceding graft failure if it occurred) (10/14 in group A and 8/13 in group B).
The incidence of CAI was 17% vs. 14% for the MMF and EC-MPS groups (not significant, Table 4). Although the overall CAI incidence was 15%, there were no protocol biopsies, just clinically indicated biopsies, performed in our study.
Comparisons of mean creatinine concentration (Cr) and estimated glomerular filtration rate (Table 5) showed similar levels in the two groups, with acceptably high renal function overall.
Patient and graft survival were similar between groups, with actuarial patient and graft survival at 48 months posttransplant being 97% and 90% in group A (six graft failures and two deaths) versus 96% and 86% in group B (eight graft failures and three deaths) (P=0.65 and 0.46, respectively). Causes of graft failure in group A were ruptured kidney (n=1), CAI (n=1), CAI/noncompliant (n=2), AR/CAI/noncompliant (n=1), and recurrent membranoproliferative glomerulonephritis (MPGN) (n=1). Causes of graft failure in group B were AR (n=1), AR/CAI (n=1), pneumonia (n=1), pneumonia/AR (n=1), AR/noncompliant (n=2), and CAI/noncompliant (n=2).
Cause of death for the single patient who died with a functioning graft in group A was malignancy (hepatocellular carcinoma). One patient in group A died after graft failure of locally advanced tongue cancer. Causes of death with a functioning graft in group B were cardiovascular event (n=1) and malignancy (n=1, metastatic tongue cancer). One patient in group B died after graft failure of a cardiovascular event (n=1).
Percentages and types of infectious complications requiring hospitalization during 48 months posttransplant were equivalent and acceptably low in the two groups (Table 6). In total, 52 patients (35%) developed an infection requiring hospitalization during the first 48 months posttransplant (23 in group A and 29 in group B); only three patients developed a CMV infection (viremia for two patients in group A and tissue invasive for one patient in group B); and none were infected by polyoma virus (BK viral titers were not monitored routinely).
Table 7 presents no differences in incidence rates of upper and lower GI adverse events between the two groups during 48 months posttransplantation. The observed percentage of patients developing any GI adverse event during 48 months posttransplantation was 45% (34/75) vs. 52% (39/75) in groups A and B, respectively (P=0.41).
Among patients having no pretransplant history of diabetes, new onset diabetes mellitus after transplantation (NODAT) developed in 13 of 61 (22%) group A patients versus 8 of 55 (15%) group B patients during the first 48 months posttransplant (P=0.39). Overall incidence of NODAT was 21 of 116 (18%). The number of NODAT patients requiring insulin/oral hypoglycemic agent was 9/4 vs. 6/2 in groups A and B, respectively. Note that 3 of 21 patients who developed NODAT were placed on maintenance steroids before development of NODAT.
Finally, there were no noteworthy differences in lipid profiles or in percentages of patients requiring antilipidemia therapy between groups (not shown).
In December 2004, we embarked on a single-center, randomized trial of 150 patients to compare administering MMF versus EC-MPS in combination with reduced tacrolimus dosing and early steroid withdrawal (7 days), where two powerful induction drugs (ATG and Dac) were used in combination to prevent early occurrence of AR (5). The reason for comparing the two IMPDH inhibitors was to determine whether the final, effective long-term immunosuppression might depend on the group that tolerated a fuller dose (presumably due to less GI toxicity) of this adjunctive agent. Although slightly closer to planned IMPDH dosages were achieved in the EC-MPS arm, there were no observed differences in GI toxicity between the two groups. Furthermore, first BPAR incidence during the first 48 months posttransplant was similarly favorable and demonstrated long-term antirejection efficacy in both study arms, with an overall first BPAR rate at 48 months of 18% (27/150), and 12% (18/150) when excluding borderline rejection episodes.
There have been few published randomized trials of MMF versus EC-MPS, two using cyclosporine microemulsion that reported similar efficacy and incidence of GI adverse events at 12 months (6, 7), and our single-center study describing 12-month (5) and now 48-month (longer term) clinical outcomes using tacrolimus in combination with steroid avoidance. The results of these three studies remain consistent, with no differences existing between the two groups in efficacy, but more importantly, in GI and other side effects. This lack of demonstrable advantage of EC-MPS over MMF is now even more relevant, as generic versions of MMF have recently become available.
The uniqueness of our study also includes the use of dual ATG/Dac induction therapy, the rationale being to combine a lymphodepleting polyclonal agent such as ATG with a (nondepleting) monoclonal antibody specifically targeting CD25 activity (and using fewer doses of each agent in comparison with either one alone). We have successfully used this combination in simultaneous kidney-pancreas transplantation (8, 9), and a recent study shows that the addition of anti-CD25 to ATG for induction therapy delays the return of peripheral blood CD25+ cells (10). Successful combination of basiliximab and ATG as dual induction in kidney transplantation has been reported (11), along with equivalency in clinical outcomes using Dac versus basiliximab (12). These results logically support the substitution of Dac (no longer available) with basiliximab. In addition, we believe that the low DGF rate at our center is explained by two factors: relatively short (static) cold storage times and the use of machine perfusion preservation for all DD kidneys (13).
In a recently published report by Ekberg et al. (4), among 401 patients who received five doses of Dac along with lower dose tacrolimus, MMF, and corticosteroids, the rejection rate at 36 months posttransplant was 14% (excluding borderline), being acceptably low and similar to our 48-month rejection rate of 12%. Renal function was also favorably high as in our study, demonstrating the effective use of reduced trough levels of tacrolimus in combination with an IMPDH inhibitor.
Although tacrolimus is known to be nephrotoxic, MMF and EC-MPS are two additionally potent noncalcineurin inhibitor immunosuppressives that are theoretically nonnephrotoxic (6, 7, 14, 15), having antiproliferative effects on vascular smooth muscle cells, perhaps also decreasing arterial intimal thickening, angiogenesis, and other less well-defined causes of CAI (15), which (in addition to AR) lead to long-term graft loss (16). Perhaps still somewhat worrisome is the small number of patients in each group that temporarily discontinued the IMPDH inhibitor (plus, others that had dose reduction), primarily due to leukopenia, infection, and GI toxicity. This might conceivably translate into a decreased protective effect against late graft loss from CAI (14).
Long-term corticosteroid therapy (CCS) is still considered by many to be part of standard maintenance immunosuppression, but multiple side effects including increases in known cardiovascular risk factors (high triglycerides, NODAT requiring insulin, and weight gain) are known. Recently, Woodle et al. (17) reported similar 5-year renal allograft survival and function in a prospective, multicenter randomized trial between early (7 day) corticosteroid withdrawal (CSWD) (n=191) versus low-dose CCS (n=195), with CSWD also providing improvements in cardiovascular risks factors. However, CSWD was associated with significantly higher BPAR (17.8%) in comparison with CCS (10.8%) (P=0.04, log-rank test). Similar results were also reported in a recent meta-analysis (18).
Promisingly, 71% (107/150) of patients in our study have been totally withdrawn from steroids throughout the first 48 months posttransplant with no apparent increase in the rate of (early and late) AR, with actuarial graft and death-censored graft survivals for the whole cohort at 48 months being 88% and 90%, respectively, perhaps also due in part to the effective combination of induction agents used here in comparison with other single antibody induction protocols previously described testing steroid withdrawal (19–21).
There was no increased frequency or severity of infections observed in either study arm during the first 48 months posttransplant. Incidence of NODAT was also acceptably low in both study arms, with an overall 18% (21/116) of patients without a pretransplant history of diabetes developing NODAT during the first 48 months posttransplant.
In conclusion, it seems that no price has been paid thus far in the attempt to reduce early- and long-term rejection episodes without increasing other adverse events through the use of rTd, full maintenance dosing with an IMPDH inhibitor, and no planned long-term corticosteroid immunosuppression therapy. Although this protocol will require (planned) longer follow-up (10 years) to demonstrate its ultimate relationship with patient and graft survival, the randomized trial component of this protocol yielded no notable differences (at 4 years posttransplant) between the MMF and EC-MPS arms.
MATERIALS AND METHODS
Between December 2004 and February 2006, 150 adult recipients (age, 18–77 years) of either DD or non-HLA identical living donor first kidney transplants were randomized in this open-labeled study immediately before transplantation: group A (n=75) received maintenance tacrolimus/MMF and group B (n=75) tacrolimus/EC-MPS. The center's institutional review board approved the protocol, and all patients gave written informed consent before enrollment. In both groups, tacrolimus was initiated at a dose of 0.1 mg/kg twice daily after renal function had improved (serum Cr decreased to <4 mg/dL absent dialysis). Target (12 hr) trough level of tacrolimus was 6 to 8 ng/mL in both groups at 12 months and 4 to 6 ng/mL thereafter. The target MMF dose in group A was 1 g twice daily; the target EC-MPS dose in group B was 720 mg twice daily. Methylprednisolone was given intravenously at 500 mg per day for 3 days postoperatively, with subsequent weaning to complete withdrawal after the first postoperative week. All patients received combined induction with rabbit ATG and Dac. ATG (1 mg/kg) was given intraoperatively, with equivalent additional postoperative doses on days 3 and 5 posttransplant. The first dose of Dac (1 mg/kg) was also given intraoperatively, with one additional dose 14 days later (5). All patients were documented as intent to treat.
The schedule of nonimmunosuppressive adjunctive therapy was the same as in our previous protocols (5, 22, 23).
Tacrolimus trough levels (12 hr) were routinely compiled for each patient, performed by whole blood immunoassay, with blood samples taken 3/wk, 2/wk, and 1/wk during the first 3 months, respectively, monthly for the next 9 months, and then once every 2 months thereafter (5). Dosing of MMF and EC-MPS was also recorded. DGF was defined as the requirement for at least one dialysis during the first week posttransplant. All patients were followed for the incidence of BPAR, graft loss, death, CAI, renal function (serum Cr and estimated glomerular filtration rate) (24), NODAT, symptomatic GI side effects, and infections requiring hospitalization.
Banff criteria were used to determine biopsy rejection severity (25). In addition, BPAR was defined as requiring both a clinical indication and treatment for the episode. Graft loss was determined as the time of reestablishment of long-term dialysis or death. NODAT was defined as the use of insulin or oral antihyperglycemic agents for ≥30 days in patients without a preoperative history of either insulin-dependent or noninsulin-dependent diabetes mellitus. Symptomatic GI adverse events were documented by clinical evaluation; dyslipidemias and statin therapy were also compiled.
Statistical analysis was performed using an intent-to-treat approach. Standard t tests were used to compare mean values between treatment arms; percentages were compared using the Pearson (uncorrected) χ2 test. Arithmetic means±standard errors were calculated except for variables that were highly skewed toward larger values, in which case geometric means and corresponding standard errors were reported, with comparisons based on log-transformed values (26). Incidences of first BPAR, NODAT, graft failure (death-censored graft loss), graft loss, and death were compared by the log-rank test, with time-to-failure curves generated using the Kaplan-Meier method. P values ≤0.05 were considered to be statistically significant.
As of the last follow-up date, March 1, 2010, all but seven patients were followed up for a minimum of 48 months posttransplant. These seven patients were lost to follow-up, with lost-to-follow-up times ranging from 13 to 48 months posttransplant (all with functioning grafts at those times). Thus, for each of the clinical outcomes, actuarial percentages of failure at 48 months posttransplant were calculated. A flow diagram for the study is presented in Figure 1.
As stated previously (5), a randomized block scheme was used in randomizing patients into the study. Specifically, patients were randomly assigned (immediately pretransplant) to one of the two treatment arms in blocks of two and four patients (block sizes were also randomly selected), ensuring a balance of patients across treatment arms after each block of patients was randomized.
1. Ekberg H, Tedesco-Silva H, Demirbas A, et al. Reduced exposure to calcineurin inhibitors in renal transplantation. N Engl J Med
2007; 357: 2562.
2. Matas AJ, Kandaswamy R, Gillingham KJ, et al. Prednisone-free maintenance immunosuppression—A 5-year experience. Am J Transplant
2005; 5: 2473.
3. Vincenti F, Larsen C, Durrbach A, et al; the belatacept study group. Costimulation blockade with belatacept in renal transplantation. N Engl J Med
2005; 353: 770.
4. Ekberg H, Bernasconi C, Tedesco-Silva H, et al. Calcineurin inhibitor minimization in the symphony study: Observational results 3 years after transplantation. Am J Transplant
2009; 9: 1876.
5. Ciancio G, Burke GW, Gaynor JJ, et al. Randomized trial of mycophenolate mofetil
versus enteric-coated mycophenolate sodium
in primary renal transplant recipients
and daclizumab/thymoglobulin: One year follow-up. Transplantation
2008; 86: 67.
6. Salvadori M, Holzer H, de Mattos A, et al. Enteric-coated mycophenolate sodium
is therapeutically equivalent to mycophenolate mofetil
in de novo renal transplant patients. Am J Transplant
2003; 4: 231.
7. Budde K, Curtis J, Knoll G, et al. Enteric-coated mycophenolate sodium
can be safely administered in maintenance renal transplant patients: Results of a 1-year study. Am J Transplant
2003; 4: 237.
8. Burke GW, Ciancio G, Figueiro J, et al. Can acute rejection be prevented in SPK transplantation? Transplant Proc
2002; 34: 1913.
9. Burke GW III, Ciancio G, Figueiro J, et al. Steroid-resistant acute rejection following SPK: Importance of maintaining therapeutic dosing in a triple-drug regimen. Transplant Proc
2002; 34: 1918.
10. Sageshima J, Ciancio G, Gaynor JJ, et al. Addition of anti-CD25 to thymoglobulin for induction therapy: Delayed return of peripheral blood CD25 positive population. Clin Transplant
11. Ruggenenti P, Codreanu I, Cravedi P, et al. Basiliximab combined with low-dose rabbit anti-human thymocyte globulin: A possible further step toward effective and minimally toxic T cell-targeted therapy in kidney transplantation. Clin J Am Soc Nephrol
2006; 1: 546.
12. Kandus A, Arnol M, Omahen K, et al. Basiliximab versus daclizumab combined with triple immunosuppression in deceased donor renal transplantation: A prospective, randomized study. Transplantation
2010; 89: 1022.
13. Ciancio G, Gaynor JJ, Sageshima J, et al. Favorable outcomes with machine perfusion and longer pump times in kidney transplantation: A single center, observational study. Transplantation
14. Ojo AO, Meier-Kriesche HU, Hanson JA, et al. Mycophenolate mofetil
reduces late renal allograft loss independent of acute rejection. Transplantation
2000; 69: 2405.
15. Gregory CR, Huang X, Pratt RE, et al. Treatment with rapamycin and mycophenolic acid reduces arterial intimal thickening produced by mechanical injury and allows endothelial replacement. Transplantation
1995; 59: 655.
16. Humar A, Hassoun A, Kandaswamy R, et al. Immunologic factors: The major risk for decreased long-term renal allograft survival. Transplantation
1999; 68: 1842.
17. Woodle ES, First MR, Pirsch J, et al. A prospective randomized, double-blind, placebo-controlled multicenter trial comparing early (7 day) corticosteroid cessation versus long-term, low-dose corticosteroid therapy. Ann Surg
2008; 248: 564.
18. Knight SR, Morris PJ. Steroid avoidance
or withdrawal after renal transplantation increases the risk of acute rejection but decreases cardiovascular risk. A meta-analysis. Transplantation
2010; 89: 1.
19. Parrott NR, Hammad AQ, Watson CJ, et al. Multicenter, randomized study of the effectiveness of basiliximab in avoiding addition of steroids to cyclosporine A monotherapy in renal transplant recipients
2005; 79: 344–348.
20. Vincenti F, Monaco A, Grinyo J, et al. Multicenter randomized prospective trial of steroid withdrawal in renal transplant recipients
receiving basiliximab, cyclosporine microemulsion and mycophenolate mofetil
. Am J Transplant
2003; 3: 306.
21. Woodle ES, Vincenti F, Lorber MI, et al. A multicenter pilot study of early (4-day) steroid cessation in renal transplant recipients
under simulect, tacrolimus
and sirolimus. Am J Transplant
2005; 5: 157.
22. Ciancio G, Burke GW, Gaynor JJ, et al. A randomized long-term trial of tacrolimus
/sirolimus versus tacrolimus
/mycophenolate versus cyclosporine/sirolimus in renal transplantation: Three-year analysis. Transplantation
2006; 81: 845.
23. Ciancio G, Burke GW, Gaynor JJ, et al. A randomized trial of three renal transplant induction antibodies: Early comparison of tacrolimus
, mycophenolate mofetil
, and steroid dosing, and newer immune-monitoring. Transplantation
2005; 80: 457.
24. Levey AS, Bosch JP, Lewis JB, et al. A more accurate method to estimate glomerular filtration rate from serum creatinine: A new prediction equation. Modification of Diet in Renal Disease Study Group. Ann Intern Med
1999; 130: 461.
25. Solez K, Colvin RB, Racusen LC, et al. Banff 07 classification of renal allograft pathology: Updates and future directions. Am J Transplant
2008; 8: 753.
26. Snedecor GW, Cochran WG. Statistical methods [ed. 6]. Ames, Iowa: The Iowa State University Press, 1967.