American Journal of Transplantation 2011; 11: 450462Wiley Periodicals Inc.
C 2010 The AuthorsJournal compilation C 2010 The American Society of
Transplantation and the American Society of Transplant Surgeons
Long-Term Renal Allograft Survival in the UnitedStates: A Critical Reappraisal
K. E. Lamb, S. Lodhi and H.-U. Meier-Kriesche
Division of Nephrology, Hypertension and Transplantation,University of Florida, Gainesville, FL*Corresponding author: Herwig-Ulf Meier-Kriesche,email@example.com
Renal allograft survival has increased tremendouslyover past decades; this has been mostly attributedto improvements in first-year survival. This report de-scribes the evolution of renal allograft survival in theUnited States where a total of 252 910 patients re-ceived a single-organ kidney transplant between 1989and 2009. Half-lives were obtained from the KaplanMeier and Cox models. Graft half-life for deceased-donor transplants was 6.6 years in 1989, increased to8 years in 1995, then after the year 2000 further in-creased to 8.8 years by 2005. More significant improve-ments were made in higher risk transplants like ECDrecipients where the half-lives increased from 3 yearsin 1989 to 6.4 years in 2005. In low-risk populationslike living-donor-recipients half-life did not change with11.4 years in 1989 and 11.9 years in 2005. First-year at-trition rates show dramatic improvements across allsubgroups; however, attrition rates beyond the firstyear show only small improvements and are some-what more evident in black recipients. The significantprogress that has occurred over the last two decadesin renal transplantation is mostly driven by improve-ments in short-term graft survival but long-term at-trition is slowly improving and could lead to biggeradvances in the future.
Key words: Graft half-life, graft survival, kidney trans-plantation
Abbreviations: SRTR, Scientific Renal Transplant Reg-istry; SCD, Standard criteria deceased donor; ECD,Expanded criteria deceased donor; Tx, Transplant;Rec/Don LE 45, Recipient and donor age less than orequal to 45; KM, Kaplan-Meier.
Received 21 June 2010, revised 13 July 2010 andaccepted for publication 04 August 2010
The specialty of kidney transplantation has made dramaticstrides over the decades evolving from an experimentalprocedure to the standard of care in the treatment of pa-tients with end-stage renal disease (1). Not only are the
outcomes after kidney transplantation good enough to im-prove the quality of life (2) of our patients, but it has alsobeen established as a life-saving procedure (3,4); yet thelife-saving benefit of a kidney transplant lasts only as longas the transplanted kidney (4). Technical and pharmaceu-tical progress have helped to improve outcomes progres-sively even over the last decade when excellent outcomeswere already considered standard of care. Now with graftsurvival rates in excess of 90% the question arises if anyfurther improvements are possible or even necessary. In2004 it became clear that the overall improvements ingraft survival after kidney transplantation were really drivenby improvements in first-year survival, whereas long-termgraft attrition remained largely unchanged over decades(5,6). This highlighted a whole other area where improve-ments might be possible and necessary. Especially nowwhen first-year survival rates are almost close to perfectit becomes clear that further improvements in long-termsurvival have to come through improvements in long graftmaintenance.
It is notoriously difficult to measure long-term survival, aslengthy follow-up is necessary to document it, yet periodicupdates on the long-term trends can potentially yield im-portant information especially when counseling patients inthe pretransplant phase regarding expectations of futureoutcomes.
The purpose of our present study was to reevaluate theevolution of short- and long-term renal allograft survival inthe United States with the most recent data provided bythe Scientific Renal Transplant Registry (SRTR).
Materials and Methods
SubjectsWe examined data from the national SRTR database for renal transplantrecipients from 1989 to November 1, 2009. Analyses were conductedon adult transplant recipients 18 years or older. Multiorgan transplantswere excluded from the analysis. Data were analyzed separately for liv-ing and deceased-donor transplants, black recipients, nonblack recipients,first transplants and repeat transplants, for standard criteria donor (SCD)kidneys and expanded criteria donor (ECD) kidneys and for recipients withdonor and recipient ages below 45 years.
Outcome measuresWe analyzed graft, patient and death-censored graft survival by estimatingsurvival half-lives and we analyzed attrition rates all stratified by year oftransplant.
Long-Term Renal Allograft Survival
Half-lives; Univariate half-lives were calculated as median half-lives, i.e.the intersection point of the KaplanMeier curve with the 50% survivalthreshold. We differentiated between actual half-lives for those instanceswhere all patients had reached the 50% mark, actuarial half-lives for thoseinstances when only a proportion of patients had reached the 50% markand projected half-lives when none of the patients reached the 50% mark.In the tables and figures actual and actuarial half-life were grouped togetherbut projected half-lives are shown separately. Projections were obtained byforecasting the KaplanMeier curves from a point of stable attrition, whichwas fairly consistently located between 3 and 8 years of survival yield-ing a period of 5 years from which the forecasts were based. Forecastedprojections were carried out using ordinary least squares point estimates.
Multivariate half-lives were obtained in the same fashion from the Coxproportional hazard models.
Attrition rates; Attrition rates were calculated by first acquiring actual 1-year,3-year, 5-year and 10-year survival rates. The total number of patients failedduring the time period was subtracted from the number of patients originallyentering the cohort and divided then by the original number entering thecohort to obtain an absolute failure percentage. The percentage of absolutefailures was then divided by the total number of years in the follow-upinterval to obtain a yearly failure rate.
Independent variablesCovariates used to calculate the adjusted half-lives from the Cox modelincluded recipients transplant age (reference group 1834), pretransplantdiagnosis of diabetes (reference group diabetic), candidate race (referencegroup Caucasian) and candidate gender (reference group males) summa-rized for transplant year 1999 to yield the most up-to-date case mix forcomplete 10-years follow-up (7).
Statistical modelsOutcomes were measured by the KaplanMeier models and the Cox mul-tivariate proportional hazard models. Half-lives were calculated based onactual and projected follow-up where applicable. Half-lives based on actualversus projected follow-up are displayed distinctly in the results. Projected
half-lives were utilized in both the univariate KaplanMeier and the multi-variate Cox Regression model for allograft failure and only in the univariateKaplanMeier for death-censored allograft failure.
Multivariate models were corrected for the same variables uses in the SRTRannual data report as described above.
Proportional hazard assumptions were tested by visually assessing loglog survival curves. The Exact method was used to handle tied outcomeoccurrences. All analyses were conducted using SAS (v.9.2, Cary, NC) anda type-one error probability of 0.05 was utilized as an indication of statisticalsignificance.
PatientsWe analyzed a total of 252 910 adult kidney recipientstransplanted between 1989 and 2005 excluding multior-gan transplants. Of these, 164 480 were deceased-donortransplants and 88 430 living-donor transplants. Of 164 480deceased-donor transplants 23 580 were ECD transplants.Of 140 900 deceased standard criteria donor recipients,120 675 were first transplant recipients and 20 225 wererepeat transplants.
Graft survivalFigure 1(A) shows overall graft survival for standard crite-ria deceased-donor transplants between 1989 and 2005and the respective median half-lives based on where thesurvival curve crosses the 50% survival line.
Figure 1(B) shows death-censored graft survival for stan-dard criteria deceased-donor transplants and the respec-tive half-lives.
Figure 1: (A) KaplanMeier cumu-lative graft failure and (B) death-censored graft failure, by yearof first deceased SCD transplantsfrom transplant year 19892008.
American Journal of Transplantation 2011; 11: 450462 451
Lamb et al.
Half-livesTable 1 displays the overall both actual or actuarial half-lives and the projected half-lives marked as forecastedin the second shaded line by transplant year. The overlapbetween the actuarial half-lives and projected half-lives rep-resents instances where still reasonable conclusions canbe drawn from the actuarial data but forecast were gener-ated in parallel. This gives a sense also about how well theforecasts might be working.
When evaluating all 164 480 deceased-donor transplantsjointly, the half-life was 6.6 years in 1989, increased toclose to 8 years in 1995, stayed around 8 years until trans-plant year 2000 and then further increased to 8.8 years in2005.
When looking only at first-donor transplants (N = 142 198)excluding retransplants, the half-life was 6.8 years in 1989,increased to 8 years in 1995 and increased to 9 years in2005.
When limiting the analysis to just standard criteriadeceased-donor transplants but including retransplants(N = 140 900) the half-lives improved from 6.7 years in1989 to 9.5 years in 2005. Slightly (