Validation of the Living Kidney Donor Profile Index in a European cohort and comparison of long-term outcomes with US results

Validation of the Living Kidney Donor Profile Index in a European cohort and comparison of... Abstract Background Recently, a risk index for living donor kidney (LDK) transplantation [living kidney donor profile index (LKDPI)] was proposed to compare LDKs with each other and with deceased donor kidneys (DDKs). Until now, the LKDPI has not been validated externally. Methods This long-term retrospective analysis included 1305 consecutive adult kidney transplant recipients who were transplanted 2000–16 in our centre. The Kidney Donor Profile Index (KDPI) was calculated in 889 DDKs and the LKDPI in 416 LDKs. Outcome was followed over a median of 6.5 years. Results The median LKDPI was 17 and the median KDPI was 69, with a high proportion of donor kidneys with a very high KDPI (40% KDPI ≥ 80). Categorization of LDK into LKDPI quartiles (LKDPI −45–3, 3–17, 17–33, 33–90) revealed a significant difference in death-censored graft survival. Comparing corresponding subgroups of the LKDPI and KDPI (LKDPI/KDPI 0–20 or 20–40) showed comparable graft survival. A multivariate analysis adjusting for relevant recipient factors revealed the KDPI [hazard ratio (HR) 1.21; P < 0.001) and LKDPI (HR 1.15; P = 0.049) as significant independent predictors of graft loss. Time-to-event receiver operating characteristic analyses for graft survival demonstrated lower predictive discrimination of the LKDPI [area under the curve (AUC) 0.55] compared with the KDPI (AUC 0.66). The 10-year graft survival of LDK recipients was inferior in the USA compared with our centre (79% versus 84%). Conclusions These results provide external validation of the LKDPI to predict death-censored graft survival and confirm comparability of the LKDPI with the KDPI to discriminate post-transplant outcome. graft survival, KDPI, LKDPI, outcome, transplantation INTRODUCTION The living donor kidney (LDK) transplantation accounts for almost 40% of all kidney transplants worldwide, expanding the limited pool of deceased donor kidneys (DDKs) [1]. Not only organ shortage, but also superior graft and patient survival compared with DDK privilege all patients with end-stage renal disease to whom a living donor is offered. However, the US living kidney donation rate steadily declined between 2004 and 2014 [2]. Recently, some improvement was reported as LDK donation rates stabilized [3]. Numerous studies have examined donor or allocation factors affecting graft or patient survival after LDK transplantations [4–6]. Recently a US-conducted study determined LDK characteristics to model all-cause graft survival and calculated a living kidney donor profile index (LKDPI) [7]. Comparable with the Kidney Donor Profile Index (KDPI) for DDKs, the LKDPI ranks LDKs based on the aggregate risk for all-cause graft loss. Additionally, the LKDPI was constructed to comparably quantify a risk level as provided by the KDPI. This allowed the study authors to compare a certain LDK with a DDK. For the LKDPI, a C-statistic of 0.59 was reported [7]. Similar C-statistics were calculated for the KDPI in DDKs [8]. Currently implementation of the KDPI in the US allocation system is under evaluation and a slightly higher discard rate through a ‘labeling effect’ for some DDKs was reported [9]. Some studies have discussed the disadvantages associated with the KDPI’s moderate discrimination strength, providing insight into the risks and benefits of a ranking metric [9–12]. Hence these analyses encourage further studies to provide a profound analysis of the LKDPI for LDKs. In addition, the LKDPI was developed in the same patient population in which it was validated and, until now, no external validation was conducted. In 2012, Ojo et al. [13] compared the long-term outcomes of DDK transplantations of a European and a US cohort. They reported striking differences in the 10-year graft survival between both cohorts (71% for European versus 53% for US recipients). Pre-transplant comorbidities and medical care of recipients were discussed as possible causes. Analysing LDK transplantations, a US-based study found some evidence that LDK recipients are overall healthier than DDK recipients and that racial and socio-economic disparities favour younger recipients and whites in the utilization of LDKs [14]. However, there is less data comparing the long-term outcomes of LDK transplantations in the USA versus Europe and analysing outcomes adjusted for LDK characteristics. The goal of this study was not only to validate the recently proposed LKDPI in another cohort but also to compare the characteristics and outcomes of LDK versus DDK recipients as well as the outcomes of US recipients versus a European cohort. MATERIALS AND METHODS This long-term retrospective analysis screened 1341 consecutive adult kidney transplantations between 2000 and 2016. In total, 416 living and 889 DDK transplantations with available donor and recipient records could be included. Complete medical records were reviewed for donor and recipient clinical and laboratory data. The characteristics were determined at the time of transplantation. Transplant outcome was followed over a maximum period of 16 years (mean 6.5). The ethics committee of Charité-Universitätsmedizin Berlin approved the study and conformity with the Declaration of Helsinki was ensured. The LKDPI was calculated as described by Massie et al. [7]. The following donor and transplant factors compose the index: age, estimated glomerular filtration rate (eGFR) [as estimated by the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) formula], body mass index (BMI), ethnicity (African American), history of cigarette use, systolic blood pressure, donor/recipient relationship, ABO incompatibility, donor/recipient sex, human leucocyte antigen (HLA) B and HLA-DR mismatches and donor:recipient weight ratio. The LKDPI was determined by mean substitution of 3.97% missing values of the aggregate values, which were used for calculation of the LKDPI in the complete cohort. A majority of LDK characteristics showed no or a low quantity of missing values regarding the LDK cohort. In particular, a minor proportion of donors had no blood group (0.2%), HLA-B or HLA-DR mismatches (both 0.5%) or GFR (7.5%) documented. Three LDK characteristics showed a slightly higher donor proportion with missing characteristics. These were BMI (11.3%), donor:recipient weight ratio (11.8%) and systolic blood pressure (15.9%). All LDK characteristics not mentioned above had no missing values. Of the donors of the LDK cohort, 78% had complete data. Notably, according to the demographics of the study population, all donors were classified as Caucasian since ethnicity is not routinely filed in Germany. Furthermore, a history of cigarette use was assumed to be negative when cigarette use was not mentioned in the individual medical record. GFR data were derived from the CKD-EPI formula (92%) or measured creatinine clearance (8%) [15]. In the case of DDK transplantations, the KDPI score for each kidney was determined using the Organ Procurement and Transplantation Network (OPTN) mapping table and the median donor of 2014 as reference [16, 17]. An aggregate of 0.45% individual missing values of all KDPI values was determined according to Organ Procurement and Transplantation Network (OPTN) methods to calculate the KDPI [17]. In detail, an unknown status of hypertension and diabetes accounted for 2.0% and 2.5%, respectively, of the DDK cohort. It should be noted that donation after cardiac death (DCD) is forbidden in Germany. All other DDK characteristics were complete. For assessment of recipient factors, the raw estimated post-transplant survival (EPTS, including recipient age, previous transplantation, recipient diabetes and time on dialysis) was calculated [18]. There were no missing data for EPTS calculation. Patients included in this study initially received a standard immunosuppressive protocol [induction therapy (anti-IL-2 R antibody], calcineurin inhibitor, mycophenolate and steroids]. If no rejection episodes occurred, tapering of steroids was performed aimed at a steroid-free regimen after the first year. Cohort characteristics and parameters were calculated by mean and standard deviation (SD) or by median and interquartile range (IQR) in case of non-parametric distributed variables. Student’s t-test or Mann–Whitney test was used for not uniformly distributed data. Chi-square tests were applied for binary variables. Survival estimates were calculated by the Kaplan–Meier method and survival differences were assessed by log-rank tests. Cox proportional hazards models were used to evaluate the prediction of graft loss. The significance level was set to α = 0.05. Subgroups were categorized by LKDPI quartiles or as proposed by Massie et al. [7] to compare LDKs with DDKs. The median eGFR was calculated by the CKD-EPI formula [15]. A GFR of 0 mL/min/1.73 m2 was imputed for patients after terminal graft failure. Graft survival was the time from transplantation to a return to dialysis, censoring for death with functioning graft. Model discrimination was analysed via time-to-event receiver operating characteristic (ROC) analysis from censored survival data using the method of Heagerty et al. [19]. The area under the curve (AUC) was calculated using data up to 10 years after transplantation. The analysis was performed using SPSS 23 for Windows (IBM, Armonk, NY, USA). Time-to-event ROC analyses were performed by R version 3.2.3 (R Project for Statistical Computing, Vienna, Austria). RESULTS This analysis included 1305 living or DDK transplantations. Baseline characteristics are shown in Table 1. As expected, LDK recipients were significantly younger, had a shorter time on dialysis, had lower proportions of prior transplantations and showed less prevalence of hypertension and diabetes than DDK recipients. Consequently, the raw EPTS of LDK recipients differed significantly from DDK recipients (1.2 versus 2.1, respectively; P < 0.001). The median age of US recipients as reported by Massie et al. [7] was higher than the age of recipients in this European cohort (LDK 50 versus 43 years, DDK 55 versus 54 years, respectively) [7]. Table 1 Donor and recipient characteristics by living and deceased donors Variable Deceased kidney donors Living kidney donors P-value n 889 416 Mean follow-up, years (SD) 6.5 (4.0) 6.5 (4.4) 0.993 Mean recipient age, years (SD) 53.5 (13.7) 43.3 (14.2) <0.001 Mean donor age, years (SD) 54.2 (15.6) 49.8 (11.5) <0.001 Recipient male, n (%) 523 (58.8) 280 (67.3) 0.003 Donor male, n (%) 481 (54.5) 152 (36.5) <0.001 Donor characteristics  Median BMI (IQR) 25.7 (23.6–27.8) 25.1 (22.5–27.8) 0.013  Median creatinine, mg/dL (IQR) 0.9 (0.7–1.2) 0.8 (0.7–0.9) <0.001  Hypertension, n (%) 347 (39.0) 91 (21.9) <0.001  Diabetes mellitus, n (%) 97 (10.9) 4 (1.0) <0.001  Median KDPI/LKDPI (IQR) 69.0 (44.0–93.0) 16.9 (3.0–33.0)  Median HLA mismatches (IQR) 3 (2–4) 3 (2–4) <0.001  Mean cold ischaemia time, hours (SD) 12.1 (5.1) 2.5 (0.8) <0.001 Recipient characteristics  Mean recipient BMI (SD) 25.6 (4.5) 25.2 (4.4) 0.083  Prior kidney transplantation, n (%) 139 (15.6) 22 (5.3) <0.001  Median time on dialysis, months (IQR) 67.0 (39.2–92.0) 10.3 (0.2–29.0) <0.001  Mean raw EPTS (SD) 2.05 (0.66) 1.20 (0.74) <0.001 Recipients’ comorbidities, n (%)  Coronary artery disease 219 (24.6) 48 (11.5) <0.001   With history of MI 71 (8.0) 13 (3.1) 0.001  Diabetes mellitus 141 (15.9) 30 (7.2) <0.001   with end-organ damage 78 (8.8) 19 (4.6) 0.006  Chronic heart failure 70 (7.9) 30 (7.2) 0.737  Peripheral arterial disease 98 (11.0) 26 (6.3) 0.006  Cerebrovascular disease 63 (7.1) 14 (3.4) 0.008  Liver disease 94 (10.6) 33 (7.9) 0.160  Peptic ulcer disease 77 (8.7) 22 (5.3) 0.033  Chronic pulmonary disease 118 (13.3) 44 (10.6) 0.178  Tumour 41 (4.6) 10 (2.4) 0.064  Connective tissue disease 17 (1.9) 14 (3.4) 0.120 Variable Deceased kidney donors Living kidney donors P-value n 889 416 Mean follow-up, years (SD) 6.5 (4.0) 6.5 (4.4) 0.993 Mean recipient age, years (SD) 53.5 (13.7) 43.3 (14.2) <0.001 Mean donor age, years (SD) 54.2 (15.6) 49.8 (11.5) <0.001 Recipient male, n (%) 523 (58.8) 280 (67.3) 0.003 Donor male, n (%) 481 (54.5) 152 (36.5) <0.001 Donor characteristics  Median BMI (IQR) 25.7 (23.6–27.8) 25.1 (22.5–27.8) 0.013  Median creatinine, mg/dL (IQR) 0.9 (0.7–1.2) 0.8 (0.7–0.9) <0.001  Hypertension, n (%) 347 (39.0) 91 (21.9) <0.001  Diabetes mellitus, n (%) 97 (10.9) 4 (1.0) <0.001  Median KDPI/LKDPI (IQR) 69.0 (44.0–93.0) 16.9 (3.0–33.0)  Median HLA mismatches (IQR) 3 (2–4) 3 (2–4) <0.001  Mean cold ischaemia time, hours (SD) 12.1 (5.1) 2.5 (0.8) <0.001 Recipient characteristics  Mean recipient BMI (SD) 25.6 (4.5) 25.2 (4.4) 0.083  Prior kidney transplantation, n (%) 139 (15.6) 22 (5.3) <0.001  Median time on dialysis, months (IQR) 67.0 (39.2–92.0) 10.3 (0.2–29.0) <0.001  Mean raw EPTS (SD) 2.05 (0.66) 1.20 (0.74) <0.001 Recipients’ comorbidities, n (%)  Coronary artery disease 219 (24.6) 48 (11.5) <0.001   With history of MI 71 (8.0) 13 (3.1) 0.001  Diabetes mellitus 141 (15.9) 30 (7.2) <0.001   with end-organ damage 78 (8.8) 19 (4.6) 0.006  Chronic heart failure 70 (7.9) 30 (7.2) 0.737  Peripheral arterial disease 98 (11.0) 26 (6.3) 0.006  Cerebrovascular disease 63 (7.1) 14 (3.4) 0.008  Liver disease 94 (10.6) 33 (7.9) 0.160  Peptic ulcer disease 77 (8.7) 22 (5.3) 0.033  Chronic pulmonary disease 118 (13.3) 44 (10.6) 0.178  Tumour 41 (4.6) 10 (2.4) 0.064  Connective tissue disease 17 (1.9) 14 (3.4) 0.120 Table 1 Donor and recipient characteristics by living and deceased donors Variable Deceased kidney donors Living kidney donors P-value n 889 416 Mean follow-up, years (SD) 6.5 (4.0) 6.5 (4.4) 0.993 Mean recipient age, years (SD) 53.5 (13.7) 43.3 (14.2) <0.001 Mean donor age, years (SD) 54.2 (15.6) 49.8 (11.5) <0.001 Recipient male, n (%) 523 (58.8) 280 (67.3) 0.003 Donor male, n (%) 481 (54.5) 152 (36.5) <0.001 Donor characteristics  Median BMI (IQR) 25.7 (23.6–27.8) 25.1 (22.5–27.8) 0.013  Median creatinine, mg/dL (IQR) 0.9 (0.7–1.2) 0.8 (0.7–0.9) <0.001  Hypertension, n (%) 347 (39.0) 91 (21.9) <0.001  Diabetes mellitus, n (%) 97 (10.9) 4 (1.0) <0.001  Median KDPI/LKDPI (IQR) 69.0 (44.0–93.0) 16.9 (3.0–33.0)  Median HLA mismatches (IQR) 3 (2–4) 3 (2–4) <0.001  Mean cold ischaemia time, hours (SD) 12.1 (5.1) 2.5 (0.8) <0.001 Recipient characteristics  Mean recipient BMI (SD) 25.6 (4.5) 25.2 (4.4) 0.083  Prior kidney transplantation, n (%) 139 (15.6) 22 (5.3) <0.001  Median time on dialysis, months (IQR) 67.0 (39.2–92.0) 10.3 (0.2–29.0) <0.001  Mean raw EPTS (SD) 2.05 (0.66) 1.20 (0.74) <0.001 Recipients’ comorbidities, n (%)  Coronary artery disease 219 (24.6) 48 (11.5) <0.001   With history of MI 71 (8.0) 13 (3.1) 0.001  Diabetes mellitus 141 (15.9) 30 (7.2) <0.001   with end-organ damage 78 (8.8) 19 (4.6) 0.006  Chronic heart failure 70 (7.9) 30 (7.2) 0.737  Peripheral arterial disease 98 (11.0) 26 (6.3) 0.006  Cerebrovascular disease 63 (7.1) 14 (3.4) 0.008  Liver disease 94 (10.6) 33 (7.9) 0.160  Peptic ulcer disease 77 (8.7) 22 (5.3) 0.033  Chronic pulmonary disease 118 (13.3) 44 (10.6) 0.178  Tumour 41 (4.6) 10 (2.4) 0.064  Connective tissue disease 17 (1.9) 14 (3.4) 0.120 Variable Deceased kidney donors Living kidney donors P-value n 889 416 Mean follow-up, years (SD) 6.5 (4.0) 6.5 (4.4) 0.993 Mean recipient age, years (SD) 53.5 (13.7) 43.3 (14.2) <0.001 Mean donor age, years (SD) 54.2 (15.6) 49.8 (11.5) <0.001 Recipient male, n (%) 523 (58.8) 280 (67.3) 0.003 Donor male, n (%) 481 (54.5) 152 (36.5) <0.001 Donor characteristics  Median BMI (IQR) 25.7 (23.6–27.8) 25.1 (22.5–27.8) 0.013  Median creatinine, mg/dL (IQR) 0.9 (0.7–1.2) 0.8 (0.7–0.9) <0.001  Hypertension, n (%) 347 (39.0) 91 (21.9) <0.001  Diabetes mellitus, n (%) 97 (10.9) 4 (1.0) <0.001  Median KDPI/LKDPI (IQR) 69.0 (44.0–93.0) 16.9 (3.0–33.0)  Median HLA mismatches (IQR) 3 (2–4) 3 (2–4) <0.001  Mean cold ischaemia time, hours (SD) 12.1 (5.1) 2.5 (0.8) <0.001 Recipient characteristics  Mean recipient BMI (SD) 25.6 (4.5) 25.2 (4.4) 0.083  Prior kidney transplantation, n (%) 139 (15.6) 22 (5.3) <0.001  Median time on dialysis, months (IQR) 67.0 (39.2–92.0) 10.3 (0.2–29.0) <0.001  Mean raw EPTS (SD) 2.05 (0.66) 1.20 (0.74) <0.001 Recipients’ comorbidities, n (%)  Coronary artery disease 219 (24.6) 48 (11.5) <0.001   With history of MI 71 (8.0) 13 (3.1) 0.001  Diabetes mellitus 141 (15.9) 30 (7.2) <0.001   with end-organ damage 78 (8.8) 19 (4.6) 0.006  Chronic heart failure 70 (7.9) 30 (7.2) 0.737  Peripheral arterial disease 98 (11.0) 26 (6.3) 0.006  Cerebrovascular disease 63 (7.1) 14 (3.4) 0.008  Liver disease 94 (10.6) 33 (7.9) 0.160  Peptic ulcer disease 77 (8.7) 22 (5.3) 0.033  Chronic pulmonary disease 118 (13.3) 44 (10.6) 0.178  Tumour 41 (4.6) 10 (2.4) 0.064  Connective tissue disease 17 (1.9) 14 (3.4) 0.120 The median LKDPI was 16.9. In all, 21.6% of all living donor grafts had a negative LKDPI (better predicted graft survival than every DDK; Figure 1). In contrast, in the US cohort, a median LKDPI of 12.8 (−0.8–27.2) and a slightly higher proportion of grafts (26.5%) with an LKDPI  < 0 were observed by Massie et al. [7]. In our cohort, transplanted DDKs had a large proportion of very high KDPI donor kidneys. This resulted in a considerably higher median KDPI of 69 (IQR 44–93) than for the kidneys recovered for transplantation in the US reference cohort in 2014 (median KDPI by definition 50) (Figure 1). FIGURE 1 View largeDownload slide Distribution of the LKDPI (living donors, n = 416) versus the KDPI (deceased donors, n = 889) of transplanted kidneys in Berlin, 2000–16. FIGURE 1 View largeDownload slide Distribution of the LKDPI (living donors, n = 416) versus the KDPI (deceased donors, n = 889) of transplanted kidneys in Berlin, 2000–16. As expected, there was a significantly better death-censored graft survival after LDK transplantations as compared with DDK transplantations (after 10 years 84 versus 70%; P < 0.001; Figure 2). The best post-transplant creatinine was 1.1 (0.9–1.3) mg/dL versus 1.1 (0.9–1.5) mg/dL (P = 0.017). FIGURE 2 View largeDownload slide Cumulative death-censored graft survival of LDK transplantation versus DDK transplantation in Berlin, 2000–16. FIGURE 2 View largeDownload slide Cumulative death-censored graft survival of LDK transplantation versus DDK transplantation in Berlin, 2000–16. Categorizing LDKs by LKDPI quartiles (LKDPI −44.5–2.9, 3.0–16.8, 17.0–33.0, 33.0–90.0) revealed a significant difference in death-censored graft survival but not in all-cause graft loss (Figure 3). Whereas LDKs of the best LKDPI quartiles (LKDPI −44.5–2.9, 3.0–16.8, 17.0–33.0) tended towards similar graft survival, graft loss occurred significantly more frequently for LDKs of the worst LKDPI quartile (LKDPI 33.0–90.0) for death-censored graft survival (after 10 years 87 versus 73%; P = 0.003) and all-cause graft loss (after 10 years 22 versus 30%; P = 0.041). FIGURE 3 View largeDownload slide Cumulative (A) death-censored graft survival and (B) living with functioning graft as categorized by LKDPI quartiles (LKDPI −45–3, 3–17, 17–33, 33–90) after LDK transplantation in Berlin, 2000–16. FIGURE 3 View largeDownload slide Cumulative (A) death-censored graft survival and (B) living with functioning graft as categorized by LKDPI quartiles (LKDPI −45–3, 3–17, 17–33, 33–90) after LDK transplantation in Berlin, 2000–16. In crude analysis, the KDPI [hazard ratio (HR) 1.15; P < 0.001], the age of the living kidney donor (HR 1.03; P = 0.046) but not the LKDPI (HR 1.11; P = 0.100) were associated with an elevated risk of death-censored graft loss (Table 2). In a multivariate analysis adjusting for age, previous transplantation, time on dialysis and recipient comorbidities (diabetes and coronary artery disease), the KDPI (HR 1.21; P < 0.001), the LKDPI (HR 1.15; P = 0.049) and the age of the living kidney donor (HR 1.03; P = 0.024) were revealed to be significant independent predictors of death-censored graft loss. However, the LKDPI as a predictor for all-cause graft loss neither reached significance in crude nor in adjusted analysis (HR 1.08; P = 0.136 and HR 1.06; P = 0.331). The KDPI’s adjusted hazard of all-cause graft loss was 1.15 per 10 increment (P < 0.001), thus slightly higher than the hazard reported by Massie et al. [7] in the US cohort (HR 1.10 per 10 increment; P < 0.001). Table 2 Univariate and multivariate Cox regression analysis of the LKDPI, KDPI and living donor age associated with an elevated risk of death-censored graft loss and all-cause graft loss in LDK and DDK transplant recipients Variable All-cause graft loss Death-censored graft loss HR (95% CI) P-value HR (95% CI) P-value Monovariate analysis  LKDPI per 10 increment (n = 416) 1.08 (0.98–1.20) 0.136 1.11 (0.98–1.27) 0.100  Living donor age (n = 416) 1.02 (1.00–1.05) 0.029 1.03 (1.00–1.06) 0.046  KDPI per 10 increment (n = 889) 1.17 (1.12–1.22) <0.001 1.15 (1.08–1.22) <0.001 Multivariate analysis  LKDPI per 10 incrementa (n = 416) 1.06 (0.95–1.18) 0.331 1.15 (1.00–1.31) 0.049  Living donor agea (n = 416) 1.02 (1.00–1.04) 0.072 1.03 (1.00–1.06) 0.024  KDPI per 10 incrementa (n = 889) 1.15 (1.09–1.21) <0.001 1.21 (1.13–1.30) <0.001 Variable All-cause graft loss Death-censored graft loss HR (95% CI) P-value HR (95% CI) P-value Monovariate analysis  LKDPI per 10 increment (n = 416) 1.08 (0.98–1.20) 0.136 1.11 (0.98–1.27) 0.100  Living donor age (n = 416) 1.02 (1.00–1.05) 0.029 1.03 (1.00–1.06) 0.046  KDPI per 10 increment (n = 889) 1.17 (1.12–1.22) <0.001 1.15 (1.08–1.22) <0.001 Multivariate analysis  LKDPI per 10 incrementa (n = 416) 1.06 (0.95–1.18) 0.331 1.15 (1.00–1.31) 0.049  Living donor agea (n = 416) 1.02 (1.00–1.04) 0.072 1.03 (1.00–1.06) 0.024  KDPI per 10 incrementa (n = 889) 1.15 (1.09–1.21) <0.001 1.21 (1.13–1.30) <0.001 a Adjusted for recipient age, time on dialysis, prior kidney transplantation, diabetes and coronary artery disease. Table 2 Univariate and multivariate Cox regression analysis of the LKDPI, KDPI and living donor age associated with an elevated risk of death-censored graft loss and all-cause graft loss in LDK and DDK transplant recipients Variable All-cause graft loss Death-censored graft loss HR (95% CI) P-value HR (95% CI) P-value Monovariate analysis  LKDPI per 10 increment (n = 416) 1.08 (0.98–1.20) 0.136 1.11 (0.98–1.27) 0.100  Living donor age (n = 416) 1.02 (1.00–1.05) 0.029 1.03 (1.00–1.06) 0.046  KDPI per 10 increment (n = 889) 1.17 (1.12–1.22) <0.001 1.15 (1.08–1.22) <0.001 Multivariate analysis  LKDPI per 10 incrementa (n = 416) 1.06 (0.95–1.18) 0.331 1.15 (1.00–1.31) 0.049  Living donor agea (n = 416) 1.02 (1.00–1.04) 0.072 1.03 (1.00–1.06) 0.024  KDPI per 10 incrementa (n = 889) 1.15 (1.09–1.21) <0.001 1.21 (1.13–1.30) <0.001 Variable All-cause graft loss Death-censored graft loss HR (95% CI) P-value HR (95% CI) P-value Monovariate analysis  LKDPI per 10 increment (n = 416) 1.08 (0.98–1.20) 0.136 1.11 (0.98–1.27) 0.100  Living donor age (n = 416) 1.02 (1.00–1.05) 0.029 1.03 (1.00–1.06) 0.046  KDPI per 10 increment (n = 889) 1.17 (1.12–1.22) <0.001 1.15 (1.08–1.22) <0.001 Multivariate analysis  LKDPI per 10 incrementa (n = 416) 1.06 (0.95–1.18) 0.331 1.15 (1.00–1.31) 0.049  Living donor agea (n = 416) 1.02 (1.00–1.04) 0.072 1.03 (1.00–1.06) 0.024  KDPI per 10 incrementa (n = 889) 1.15 (1.09–1.21) <0.001 1.21 (1.13–1.30) <0.001 a Adjusted for recipient age, time on dialysis, prior kidney transplantation, diabetes and coronary artery disease. Model discrimination was superior for the KDPI compared with the LKDPI. The predictive strength for death-censored graft survival or living with a functioning graft investigated by the AUC was 0.55 and 0.53 for the LKDPI and 0.66 and 0.65 for the KDPI , respectively (Figure 4). As there was limited predictive strength investigating the AUC of a continuous LKDPI, discrimination was examined in a dichotomous way (LKDPI  <33 versus LKDPI  ≥33) and hence a better discrimination was obtained (AUC of 0.60 for death-censored graft survival and AUC of 0.54 for living with a functioning graft). Furthermore, these results were qualitatively reproduced in a subgroup of 323 LDK recipients who had complete data for calculation of the LKDPI (no missing values except ethnicity, which is not routinely filed in Germany). Although limited due to the smaller sample size and shorter follow-up, the discriminative ability of the LKDPI was comparable and the C-statistic resulted in an AUC of 0.58 for death-censored graft loss and 0.52 for all-cause graft loss up to 8 years post-transplant in this subgroup. By calculating the discrimination of a refit model, Massie et al. [7] reported a C-statistic of 0.59 for the LKDPI and 0.58 for the KDPI. FIGURE 4 View largeDownload slide Time-to-event ROC curves for the LKDPI (n = 416) and KDPI (n = 889) for (A) death-censored graft survival and (B) living with functioning graft. FIGURE 4 View largeDownload slide Time-to-event ROC curves for the LKDPI (n = 416) and KDPI (n = 889) for (A) death-censored graft survival and (B) living with functioning graft. In subgroups with corresponding LKDPI/KDPI (0–20, 20–40), we compared the incidence of death-censored graft loss and all-cause graft loss of LDKs versus DDKs. Results were comparable and did not differ significantly [death-censored graft survival P = 0.704 and P = 0.711 (Figure 5); all-cause graft loss P = 0.231 and P = 0.949]. FIGURE 5 View largeDownload slide Recipients' cumulative death-censored graft survival after LDK transplantation as categorized by the LKDPI versus DDK transplantation as categorized by the KDPI: (A) LKDPI/KDPI 0–20, (B) 20–40. FIGURE 5 View largeDownload slide Recipients' cumulative death-censored graft survival after LDK transplantation as categorized by the LKDPI versus DDK transplantation as categorized by the KDPI: (A) LKDPI/KDPI 0–20, (B) 20–40. Comparing the imputed eGFR of the complete cohort showed a significantly better eGFR for LDK recipients than for DDK recipients at all time points (Figure 6). Nevertheless, the comparison of the eGFR in subgroups with corresponding LKDPI/KDPI revealed no significant differences in the imputed eGFR of 49 versus 56 mL/min/1.73 m2 (P = 0.187, LKDPI/KDPI 0–20) and versus 44 mL/min/1.73 m2 (P = 0.707, LKDPI/KDPI 20–40) at 8 years post-transplant. Interestingly, the 10-year survival of LDK and DDK recipients also proved to be comparable [LKDPI/KDPI 0–20: 86% and 81%; P = 0.331 and LKDPI/KDPI 20–40: 89% (both); P = 0.449]. LDK transplantations did not contribute an independent benefit to improve death-censored graft survival in a multivariate analysis adjusted for recipient age, time on dialysis, prior kidney transplantation, diabetes and coronary artery disease (LKDPI/KDPI 0–20: HR 1.40, P = 0.480; 20–40: HR 1.79, P = 0.325). FIGURE 6 View largeDownload slide Median eGFR using imputation for values after graft loss (patients with graft loss: GFR = 0 mL/min/1.73 m2). eGFR was calculated using the CKD-EPI formula after LDK transplantation versus DDK transplantation: (A) all recipients, (B) LKDPI/KDPI 0–20, (C) LKDPI/KDPI 20–40. FIGURE 6 View largeDownload slide Median eGFR using imputation for values after graft loss (patients with graft loss: GFR = 0 mL/min/1.73 m2). eGFR was calculated using the CKD-EPI formula after LDK transplantation versus DDK transplantation: (A) all recipients, (B) LKDPI/KDPI 0–20, (C) LKDPI/KDPI 20–40. The aggregate cumulative incidence of all-cause graft loss after 8 years was roughly 20% for American LDK recipients, who received an LDK with an LKDPI  < 0, and roughly 40% for recipients who received an LDK with an LKDPI >40 [20]. At our transplant centre, all-cause graft loss after 8 years occurred less frequently (12% and 24%, respectively). DISCUSSION For the first time, we provide external validation of the recently proposed LKDPI in a non-American cohort. Although the choice of LDK transplantation is highly driven by the living donor’s risk of end-stage renal disease, it may be valuable to assess the donor kidney quality in order to compare LDKs with each other and with DDKs [21]. LDK quality as defined by the LKDPI was slightly better in the US cohort analysed by Massie et al. [7] than in this European cohort (LKDPI 12.8 versus 16.9). The US allocation system implemented the KDPI for DDKs in late 2014. Its purpose is to match the longevity of the donor kidney with the estimated survival of the recipient as defined by the EPTS score [8]. In addition, the dichotomous prior allocation routine was thought to complicate the utilization of marginal donor grafts. The KDPI is also a useful tool to compare DDK quality recovered in different countries. A high proportion of very high KDPI kidneys allocated within the Eurotransplant system were transplanted at our centre, resulting in a median KDPI considerably higher than in the US cohort. In 2013, less than one-tenth (9%) of DDKs recovered for transplantation by the US allocation system had a KDPI >85, whereas 35% of all DDK recipients of our cohort received such a graft. Moreover, these very high KDPI kidneys have a high discard rate (55%) in the USA [22]. Since the KDPI is not implemented within the Eurotransplant system, there are few data about KDPI-assessed DDK quality transplanted within other Eurotransplant member countries. It is likely that differences may be expected as donation rates between Eurotransplant countries and policies between transplant centres show divergences. A prospective single-centre study at the Antwerp University Hospital in Belgium calculated a considerably lower median KDPI [47 (IQR 33–58)] with an inferior proportion of DDKs with a KDPI >85 (1.3%) compared with our DDK cohort [23]. However, a combined retro- and prospective multicentre study from Italy included 248 standard criteria donors [mean KDPI 35 (SD 19.2)] and 442 marginal donors (≥65 years or renal dysfunction) and thus received a higher KDPI for the whole cohort [66.6 (SD 29.6)] with a high proportion of high KDPI kidneys (30% KDPI  ≥91) [10]. Of note, all consecutive DDK transplantations in our centre were included and the proportion of donors ≥ 65 years was 30%. Finally, a retrospective multicentre study from The Netherlands revealed a median KDRI of 1.21, comparable with a median KDRI of 1.24 in the USA in 2012 [24]. Encouraged by promising follow-up data, our transplant centre pursues a guideline-directed selection of the LDK and DDK donor aimed at a good functional match. Moreover, especially in patients who received marginal organs, the transplant outcome depends on close post-transplant monitoring. As expected, LDK recipients had a significantly better graft survival than DDK recipients in this cohort despite an overlap between LDK and DDK quality as shown by Massie et al. [7] and confirmed in this cohort. There are relevant differences comparing death-censored graft survival in the USA and in our cohort. For instance, LDKs transplanted in 2003 in the USA showed a 10-year death-censored graft survival of 79 versus 84% in our cohort [22]. Comparably, although a higher proportion of very high KDPI kidneys was transplanted at our centre, US recipients of DDKs when transplanted in 2003 showed a similar death-censored graft survival compared with DDK recipients at our centre (after 10 years, 72 versus 70%) [22]. Whether different post-transplant care and insurance policies (especially as parts of US Medicare ensure immunosuppressive drug coverage only up to 36 months post-transplant) could explain these survival disparities to some extent is a question beyond the scope of this study [13, 25]. Previous studies reported higher mortality as well as worse graft survival in recipients showing multiple comorbidities [26–28]. In this regard, confounding recipient characteristics should be considered to assess the LKDPI’s performance. Although LDK recipients at our centre were significantly younger and showed less prevalence of comorbidities than DDK recipients, a relevant proportion had important comorbidities such as coronary artery disease or diabetes mellitus. Indeed, an adjusted analysis for recipients’ comorbidities and allocation factors confirmed the LKDPI as a significant and independent risk factor of death-censored graft survival in our cohort of 416 LDK recipients (Table 2). The LKDPI demonstrated only moderate discrimination in this cohort (C-statistic of 0.55 for death-censored graft survival). Massie et al. [7] hypothesized that the range of quality is narrower for LDK than for DDK donors, especially as DDK donors show a broader spectrum of comorbidities. Notably, the IQR as an indicator of the range of quality was also narrower for LDKs of our cohort (LDK 30 versus DDK 49). Schold et al. [20] addressed the limited predictive power of the LKDPI and advised cautious application to individual decision-making. Probably the LKDPI’s low discrimination may simply emphasize the unneglectable effect of recipient factors on graft and patient survival [7]. Further investigation in other cohorts may be necessary to better investigate the LKDPI’s ability to guide the individual patient. In our cohort, the specificity and sensitivity of the LKDPI tended to be higher for high LKDPI kidneys (Figure 4), thus suggesting further analysis to examine the LKDPI as a dichotomous variable. As our data indicate a survival disadvantage in particular for LKDPI kidneys in the worst LKDPI quartile, we analysed the predictive strength of LKDPI  ≥33 versus LKDPI <33. Indeed, a considerably higher predictive power results for death-censored graft survival. In a dichotomous view, LKDPI  ≥33 is also a strong and significant risk factor of death-censored graft survival (adjusted HR 3.01; P = 0.001). Regarding DDK transplantation in the USA, there is a dichotomous use of the KDPI in several cases. For example, high KDPI kidneys (KDPI >85) are designated to patients who give a candidate consent and the best KDPI kidneys are assigned to recipients with an EPTS  ≤20. However, to evaluate how this practice could apply to the LKDPI exceeds the scope of this study. Our findings are in line with prior reported data of an increasing C-statistic in high KDRI transplants [29]. The discrimination of the KDPI derived from DDKs transplanted at our centre was similar to previously published data [8]. Massie et al. [7] reported higher donor age among living donors >50 years of age as a significant predictor of reduced all-cause graft loss. Living donor age was also investigated by Noppakun et al. [5]. They reported an association between increasing donor age and worse outcome that was observed particularly after 4 years post-transplant and modified by recipient age. Furthermore, several other studies reported donor age as a significant predictor of reduced graft survival when used as a dichotomous parameter [6, 30, 31]. We investigated older living donor age as a continuous predictor of death-censored graft loss in an adjusted analysis, confirming a significant predictive value with an HR of 1.03 per year. Thus our data emphasize the great influence of increasing age of the living kidney donor regardless of the age percentile. Additionally, the hazard of increasing age of the living kidney donor increases when calculated in the subgroup of donors >50 years of age (HR 1.17; P < 0.001). This is of particular interest as since, on the one hand, there is a very high proportion of living donors >50 years of age in our cohort (50%) and, on the other hand, the age of living kidney donors at transplantation generally increases [3]. Massie et al. [7] developed the LKDPI for comparison of LDKs with DDKs. As intended by the LKDPI algorithm, our analysis of LDKs also showed similar graft survival and post-transplant eGFR compared with DDKs in subgroups of corresponding LKDPI/KDPI. Further analysis revealed comparable patient survival and no significant benefit of an LDK donation according to an adjusted multivariate analysis taking relevant recipient factors into account. Since the LKDPI is able to significantly stratify the risk of death-censored graft loss, these findings indicate comparability of the LKDPI and KDPI in a certain range of kidney quality. By performing an observational study, we could only rely on data recorded at our centre. Some data were missing in a minority of living donors. For example, since ethnic information is not routinely recorded in the Eurotransplant system, for calculation of the LKDPI, all patients were classified as Caucasian according to the population structure in Germany. While the original study excluded previously transplanted patients, we included these patients and adjusted for this characteristic in a multivariate Cox analysis. These results provide external validation of the LKDPI as a tool to assess LDK quality and prognosis of graft survival. The LKDPI may be valuable to compare LDKs with each other and with DDKs, although the moderate discriminative ability indicates limited clinical use for guidance of the individual patient. Comparing the outcome of this cohort with US data, a worse death-censored graft survival for LDK recipients in the USA is shown. In addition, comparing corresponding subgroups of LDKs (LKDPI  <0 or LKDPI >40) revealed a higher all-cause graft loss at 8 years post-transplant in the USA. AUTHORS’ CONTRIBUTIONS O.S., G.R., A.K., A.S. and M.D. were responsible for the conception and design or analysis and interpretation of data, or both. G.R., O.S., F.H., D.K., L.L. and K.B. were responsible for drafting the article or revising it. G.R., O.S., K.B., F.H., D.K., L.L. and M.D. provided intellectual content of critical importance to the work described. G.R., F.H., D.K., L.L, A.K., A.S., M.D., K.B. and O.S. approved the final version to be published. CONFLICT OF INTEREST STATEMENT None declared. The results presented in this article have not been published previously in whole or part, except in abstract format. REFERENCES 1 Cozzi E , Biancone L , Lopez-Fraga M et al. Long-term outcome of living kidney donation: position paper of the European Committee on Organ Transplantation, Council of Europe . Transplantation 2016 ; 100 : 270 – 271 Google Scholar CrossRef Search ADS PubMed 2 Hart A , Smith JM , Skeans MA et al. Kidney . Am J Transplant 2016 ; 16 : 11 – 46 Google Scholar CrossRef Search ADS PubMed 3 Hart A , Smith JM , Skeans MA et al. OPTN/SRTR 2015 annual data report: kidney . Am J Transplant 2017 ; 17 : 21 – 116 Google Scholar CrossRef Search ADS PubMed 4 Araujo AM , Santos F , Guimaraes J et al. Living-donor kidney transplantation: predictive factors and impact on post-transplant outcome . Transplant Proc 2015 ; 47 : 938 – 941 Google Scholar CrossRef Search ADS PubMed 5 Noppakun K , Cosio FG , Dean PG et al. Living donor age and kidney transplant outcomes . Am J Transplant 2011 ; 11 : 1279 – 1286 Google Scholar CrossRef Search ADS PubMed 6 Fuggle SV , Allen JE , Johnson RJ et al. Factors affecting graft and patient survival after live donor kidney transplantation in the UK . Transplantation 2010 ; 89 : 694 – 701 Google Scholar CrossRef Search ADS PubMed 7 Massie AB , Leanza J , Fahmy LM et al. A risk index for living donor kidney transplantation . Am J Transplant 2016 ; 16 : 2077 – 2084 Google Scholar CrossRef Search ADS PubMed 8 Lee AP , Abramowicz D. Is the Kidney Donor Risk Index a step forward in the assessment of deceased donor kidney quality? Nephrol Dial Transplant 2015 ; 30 : 1285 – 1290 Google Scholar CrossRef Search ADS PubMed 9 Bae S , Massie AB , Luo X et al. Changes in discard rate after the introduction of the Kidney Donor Profile Index (KDPI) . Am J Transplant 2016 ; 16 : 2202 – 2207 Google Scholar CrossRef Search ADS PubMed 10 Gandolfini I , Buzio C , Zanelli P et al. The Kidney Donor Profile Index (KDPI) of marginal donors allocated by standardized pretransplant donor biopsy assessment: distribution and association with graft outcomes . Am J Transplant 2014 ; 14 : 2515 – 2525 Google Scholar CrossRef Search ADS PubMed 11 Tanriover B , Mohan S , Cohen DJ et al. Kidneys at higher risk of discard: expanding the role of dual kidney transplantation . Am J Transplant 2014 ; 14 : 404 – 415 Google Scholar CrossRef Search ADS PubMed 12 Gupta A , Chen G , Kaplan B. KDPI and donor selection . Am J Transplant 2014 ; 14 : 2444 – 2445 Google Scholar CrossRef Search ADS PubMed 13 Ojo AO , Morales JM , Gonzalez-Molina M et al. Comparison of the long-term outcomes of kidney transplantation: USA versus Spain . Nephrol Dial Transplant 2013 ; 28 : 213 – 220 Google Scholar CrossRef Search ADS PubMed 14 Gore JL , Danovitch GM , Litwin MS et al. Disparities in the utilization of live donor renal transplantation . Am J Transplant 2009 ; 9 : 1124 – 1133 Google Scholar CrossRef Search ADS PubMed 15 Levey AS , Stevens LA , Schmid CH et al. A new equation to estimate glomerular filtration rate . Ann Intern Med 2009 ; 150 : 604 – 612 Google Scholar CrossRef Search ADS PubMed 16 Rao PS , Schaubel DE , Guidinger MK et al. A comprehensive risk quantification score for deceased donor kidneys: the kidney donor risk index . Transplantation 2009 ; 88 : 231 – 236 Google Scholar CrossRef Search ADS PubMed 17 OPTN/UNOS . A Guide to Calculating and Interpretating the Kidney Donor Profile Index (KDPI). Richmond, VA: Organ Procurement and Transplantation Network, 2016. Available at: https://optn.transplant.hrsa.gov/media/1512/guide_to_calculating_interpreting_kdpi.pdf (25 April 2018, date last accessed) 18 OPTN . A Guide to Calculating and Interpretating the Estimated Post-Transplant Survival (EPTS) Score Used in the Kidney Allocation System (KAS). Richmond, VA: Organ Procurement and Transplantation Network, 2014. Available at: https://optn.transplant.hrsa.gov/media/1511/guide_to_calculating_interpreting_epts.pdf (25 April 2018, date last accessed) 19 Heagerty PJ , Lumley T , Pepe MS. Time-dependent ROC curves for censored survival data and a diagnostic marker . Biometrics 2000 ; 56 : 337 – 344 Google Scholar CrossRef Search ADS PubMed 20 Schold JD , Kaplan B. Living kidney donor profile index: utility and limitations . Am J Transplant 2016 ; 16 : 1951 – 1952 Google Scholar CrossRef Search ADS PubMed 21 Grams ME , Garg AX , Lentine KL. Kidney-failure risk projection for the living kidney-donor candidate . N Engl J Med 2016 ; 374 : 2094 – 2095 Google Scholar CrossRef Search ADS PubMed 22 Matas AJ , Smith JM , Skeans MA. OPTN/SRTR 2013 annual data report: kidney . Am J Transplant 2015 ; 15 (Suppl 2): 1 – 34 Google Scholar CrossRef Search ADS PubMed 23 Philipse E , Lee APK , Bracke B et al. Does Kidney Donor Risk Index implementation lead to the transplantation of more and higher-quality donor kidneys? Nephrol Dial Transplant 2017 ; 32 : 1934 – 1938 Google Scholar CrossRef Search ADS PubMed 24 Peters-Sengers H , Heemskerk MBA , Geskus RB et al. Validation of the prognostic kidney donor risk index scoring system of deceased donors for renal transplantation in the Netherlands . Transplantation 2018 ; 102 : 162 – 170 Google Scholar CrossRef Search ADS PubMed 25 Tanriover B , Stone PW , Mohan S et al. Future of Medicare immunosuppressive drug coverage for kidney transplant recipients in the United States . Clin J Am Soc Nephrol 2013 ; 8 : 1258 – 1266 Google Scholar CrossRef Search ADS PubMed 26 Jassal SV , Schaubel DE , Fenton SS. Baseline comorbidity in kidney transplant recipients: a comparison of comorbidity indices . Am J Kidney Dis 2005 ; 46 : 136 – 142 Google Scholar CrossRef Search ADS PubMed 27 Wu C , Evans I , Joseph R et al. Comorbid conditions in kidney transplantation: association with graft and patient survival . J Am Soc Nephrol 2005 ; 16 : 3437 – 3444 Google Scholar CrossRef Search ADS PubMed 28 Halleck F , Khadzhynov D , Lehner L et al. Impact of pre-existing comorbidities on long-term patient and graft survival in kidney transplant recipients . Am J Transplant 2016 ; 16 (Suppl 3): 489 – 489 29 Akkina SK , Asrani SK , Peng Y et al. Development of organ-specific donor risk indices . Liver Transpl 2012 ; 18 : 395 – 404 Google Scholar CrossRef Search ADS PubMed 30 Issa N , Stephany B , Fatica R et al. Donor factors influencing graft outcomes in live donor kidney transplantation . Transplantation 2007 ; 83 : 593 – 599 Google Scholar CrossRef Search ADS PubMed 31 Young A , Kim SJ , Speechley MR et al. Accepting kidneys from older living donors: impact on transplant recipient outcomes . Am J Transplant 2011 ; 11 : 743 – 750 Google Scholar CrossRef Search ADS PubMed © The Author(s) 2018. Published by Oxford University Press on behalf of ERA-EDTA. All rights reserved. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Nephrology Dialysis Transplantation Oxford University Press

Validation of the Living Kidney Donor Profile Index in a European cohort and comparison of long-term outcomes with US results

Loading next page...
 
/lp/ou_press/validation-of-the-living-kidney-donor-profile-index-in-a-european-H1io2vhiCE
Publisher
Oxford University Press
Copyright
© The Author(s) 2018. Published by Oxford University Press on behalf of ERA-EDTA. All rights reserved.
ISSN
0931-0509
eISSN
1460-2385
D.O.I.
10.1093/ndt/gfy118
Publisher site
See Article on Publisher Site

Abstract

Abstract Background Recently, a risk index for living donor kidney (LDK) transplantation [living kidney donor profile index (LKDPI)] was proposed to compare LDKs with each other and with deceased donor kidneys (DDKs). Until now, the LKDPI has not been validated externally. Methods This long-term retrospective analysis included 1305 consecutive adult kidney transplant recipients who were transplanted 2000–16 in our centre. The Kidney Donor Profile Index (KDPI) was calculated in 889 DDKs and the LKDPI in 416 LDKs. Outcome was followed over a median of 6.5 years. Results The median LKDPI was 17 and the median KDPI was 69, with a high proportion of donor kidneys with a very high KDPI (40% KDPI ≥ 80). Categorization of LDK into LKDPI quartiles (LKDPI −45–3, 3–17, 17–33, 33–90) revealed a significant difference in death-censored graft survival. Comparing corresponding subgroups of the LKDPI and KDPI (LKDPI/KDPI 0–20 or 20–40) showed comparable graft survival. A multivariate analysis adjusting for relevant recipient factors revealed the KDPI [hazard ratio (HR) 1.21; P < 0.001) and LKDPI (HR 1.15; P = 0.049) as significant independent predictors of graft loss. Time-to-event receiver operating characteristic analyses for graft survival demonstrated lower predictive discrimination of the LKDPI [area under the curve (AUC) 0.55] compared with the KDPI (AUC 0.66). The 10-year graft survival of LDK recipients was inferior in the USA compared with our centre (79% versus 84%). Conclusions These results provide external validation of the LKDPI to predict death-censored graft survival and confirm comparability of the LKDPI with the KDPI to discriminate post-transplant outcome. graft survival, KDPI, LKDPI, outcome, transplantation INTRODUCTION The living donor kidney (LDK) transplantation accounts for almost 40% of all kidney transplants worldwide, expanding the limited pool of deceased donor kidneys (DDKs) [1]. Not only organ shortage, but also superior graft and patient survival compared with DDK privilege all patients with end-stage renal disease to whom a living donor is offered. However, the US living kidney donation rate steadily declined between 2004 and 2014 [2]. Recently, some improvement was reported as LDK donation rates stabilized [3]. Numerous studies have examined donor or allocation factors affecting graft or patient survival after LDK transplantations [4–6]. Recently a US-conducted study determined LDK characteristics to model all-cause graft survival and calculated a living kidney donor profile index (LKDPI) [7]. Comparable with the Kidney Donor Profile Index (KDPI) for DDKs, the LKDPI ranks LDKs based on the aggregate risk for all-cause graft loss. Additionally, the LKDPI was constructed to comparably quantify a risk level as provided by the KDPI. This allowed the study authors to compare a certain LDK with a DDK. For the LKDPI, a C-statistic of 0.59 was reported [7]. Similar C-statistics were calculated for the KDPI in DDKs [8]. Currently implementation of the KDPI in the US allocation system is under evaluation and a slightly higher discard rate through a ‘labeling effect’ for some DDKs was reported [9]. Some studies have discussed the disadvantages associated with the KDPI’s moderate discrimination strength, providing insight into the risks and benefits of a ranking metric [9–12]. Hence these analyses encourage further studies to provide a profound analysis of the LKDPI for LDKs. In addition, the LKDPI was developed in the same patient population in which it was validated and, until now, no external validation was conducted. In 2012, Ojo et al. [13] compared the long-term outcomes of DDK transplantations of a European and a US cohort. They reported striking differences in the 10-year graft survival between both cohorts (71% for European versus 53% for US recipients). Pre-transplant comorbidities and medical care of recipients were discussed as possible causes. Analysing LDK transplantations, a US-based study found some evidence that LDK recipients are overall healthier than DDK recipients and that racial and socio-economic disparities favour younger recipients and whites in the utilization of LDKs [14]. However, there is less data comparing the long-term outcomes of LDK transplantations in the USA versus Europe and analysing outcomes adjusted for LDK characteristics. The goal of this study was not only to validate the recently proposed LKDPI in another cohort but also to compare the characteristics and outcomes of LDK versus DDK recipients as well as the outcomes of US recipients versus a European cohort. MATERIALS AND METHODS This long-term retrospective analysis screened 1341 consecutive adult kidney transplantations between 2000 and 2016. In total, 416 living and 889 DDK transplantations with available donor and recipient records could be included. Complete medical records were reviewed for donor and recipient clinical and laboratory data. The characteristics were determined at the time of transplantation. Transplant outcome was followed over a maximum period of 16 years (mean 6.5). The ethics committee of Charité-Universitätsmedizin Berlin approved the study and conformity with the Declaration of Helsinki was ensured. The LKDPI was calculated as described by Massie et al. [7]. The following donor and transplant factors compose the index: age, estimated glomerular filtration rate (eGFR) [as estimated by the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) formula], body mass index (BMI), ethnicity (African American), history of cigarette use, systolic blood pressure, donor/recipient relationship, ABO incompatibility, donor/recipient sex, human leucocyte antigen (HLA) B and HLA-DR mismatches and donor:recipient weight ratio. The LKDPI was determined by mean substitution of 3.97% missing values of the aggregate values, which were used for calculation of the LKDPI in the complete cohort. A majority of LDK characteristics showed no or a low quantity of missing values regarding the LDK cohort. In particular, a minor proportion of donors had no blood group (0.2%), HLA-B or HLA-DR mismatches (both 0.5%) or GFR (7.5%) documented. Three LDK characteristics showed a slightly higher donor proportion with missing characteristics. These were BMI (11.3%), donor:recipient weight ratio (11.8%) and systolic blood pressure (15.9%). All LDK characteristics not mentioned above had no missing values. Of the donors of the LDK cohort, 78% had complete data. Notably, according to the demographics of the study population, all donors were classified as Caucasian since ethnicity is not routinely filed in Germany. Furthermore, a history of cigarette use was assumed to be negative when cigarette use was not mentioned in the individual medical record. GFR data were derived from the CKD-EPI formula (92%) or measured creatinine clearance (8%) [15]. In the case of DDK transplantations, the KDPI score for each kidney was determined using the Organ Procurement and Transplantation Network (OPTN) mapping table and the median donor of 2014 as reference [16, 17]. An aggregate of 0.45% individual missing values of all KDPI values was determined according to Organ Procurement and Transplantation Network (OPTN) methods to calculate the KDPI [17]. In detail, an unknown status of hypertension and diabetes accounted for 2.0% and 2.5%, respectively, of the DDK cohort. It should be noted that donation after cardiac death (DCD) is forbidden in Germany. All other DDK characteristics were complete. For assessment of recipient factors, the raw estimated post-transplant survival (EPTS, including recipient age, previous transplantation, recipient diabetes and time on dialysis) was calculated [18]. There were no missing data for EPTS calculation. Patients included in this study initially received a standard immunosuppressive protocol [induction therapy (anti-IL-2 R antibody], calcineurin inhibitor, mycophenolate and steroids]. If no rejection episodes occurred, tapering of steroids was performed aimed at a steroid-free regimen after the first year. Cohort characteristics and parameters were calculated by mean and standard deviation (SD) or by median and interquartile range (IQR) in case of non-parametric distributed variables. Student’s t-test or Mann–Whitney test was used for not uniformly distributed data. Chi-square tests were applied for binary variables. Survival estimates were calculated by the Kaplan–Meier method and survival differences were assessed by log-rank tests. Cox proportional hazards models were used to evaluate the prediction of graft loss. The significance level was set to α = 0.05. Subgroups were categorized by LKDPI quartiles or as proposed by Massie et al. [7] to compare LDKs with DDKs. The median eGFR was calculated by the CKD-EPI formula [15]. A GFR of 0 mL/min/1.73 m2 was imputed for patients after terminal graft failure. Graft survival was the time from transplantation to a return to dialysis, censoring for death with functioning graft. Model discrimination was analysed via time-to-event receiver operating characteristic (ROC) analysis from censored survival data using the method of Heagerty et al. [19]. The area under the curve (AUC) was calculated using data up to 10 years after transplantation. The analysis was performed using SPSS 23 for Windows (IBM, Armonk, NY, USA). Time-to-event ROC analyses were performed by R version 3.2.3 (R Project for Statistical Computing, Vienna, Austria). RESULTS This analysis included 1305 living or DDK transplantations. Baseline characteristics are shown in Table 1. As expected, LDK recipients were significantly younger, had a shorter time on dialysis, had lower proportions of prior transplantations and showed less prevalence of hypertension and diabetes than DDK recipients. Consequently, the raw EPTS of LDK recipients differed significantly from DDK recipients (1.2 versus 2.1, respectively; P < 0.001). The median age of US recipients as reported by Massie et al. [7] was higher than the age of recipients in this European cohort (LDK 50 versus 43 years, DDK 55 versus 54 years, respectively) [7]. Table 1 Donor and recipient characteristics by living and deceased donors Variable Deceased kidney donors Living kidney donors P-value n 889 416 Mean follow-up, years (SD) 6.5 (4.0) 6.5 (4.4) 0.993 Mean recipient age, years (SD) 53.5 (13.7) 43.3 (14.2) <0.001 Mean donor age, years (SD) 54.2 (15.6) 49.8 (11.5) <0.001 Recipient male, n (%) 523 (58.8) 280 (67.3) 0.003 Donor male, n (%) 481 (54.5) 152 (36.5) <0.001 Donor characteristics  Median BMI (IQR) 25.7 (23.6–27.8) 25.1 (22.5–27.8) 0.013  Median creatinine, mg/dL (IQR) 0.9 (0.7–1.2) 0.8 (0.7–0.9) <0.001  Hypertension, n (%) 347 (39.0) 91 (21.9) <0.001  Diabetes mellitus, n (%) 97 (10.9) 4 (1.0) <0.001  Median KDPI/LKDPI (IQR) 69.0 (44.0–93.0) 16.9 (3.0–33.0)  Median HLA mismatches (IQR) 3 (2–4) 3 (2–4) <0.001  Mean cold ischaemia time, hours (SD) 12.1 (5.1) 2.5 (0.8) <0.001 Recipient characteristics  Mean recipient BMI (SD) 25.6 (4.5) 25.2 (4.4) 0.083  Prior kidney transplantation, n (%) 139 (15.6) 22 (5.3) <0.001  Median time on dialysis, months (IQR) 67.0 (39.2–92.0) 10.3 (0.2–29.0) <0.001  Mean raw EPTS (SD) 2.05 (0.66) 1.20 (0.74) <0.001 Recipients’ comorbidities, n (%)  Coronary artery disease 219 (24.6) 48 (11.5) <0.001   With history of MI 71 (8.0) 13 (3.1) 0.001  Diabetes mellitus 141 (15.9) 30 (7.2) <0.001   with end-organ damage 78 (8.8) 19 (4.6) 0.006  Chronic heart failure 70 (7.9) 30 (7.2) 0.737  Peripheral arterial disease 98 (11.0) 26 (6.3) 0.006  Cerebrovascular disease 63 (7.1) 14 (3.4) 0.008  Liver disease 94 (10.6) 33 (7.9) 0.160  Peptic ulcer disease 77 (8.7) 22 (5.3) 0.033  Chronic pulmonary disease 118 (13.3) 44 (10.6) 0.178  Tumour 41 (4.6) 10 (2.4) 0.064  Connective tissue disease 17 (1.9) 14 (3.4) 0.120 Variable Deceased kidney donors Living kidney donors P-value n 889 416 Mean follow-up, years (SD) 6.5 (4.0) 6.5 (4.4) 0.993 Mean recipient age, years (SD) 53.5 (13.7) 43.3 (14.2) <0.001 Mean donor age, years (SD) 54.2 (15.6) 49.8 (11.5) <0.001 Recipient male, n (%) 523 (58.8) 280 (67.3) 0.003 Donor male, n (%) 481 (54.5) 152 (36.5) <0.001 Donor characteristics  Median BMI (IQR) 25.7 (23.6–27.8) 25.1 (22.5–27.8) 0.013  Median creatinine, mg/dL (IQR) 0.9 (0.7–1.2) 0.8 (0.7–0.9) <0.001  Hypertension, n (%) 347 (39.0) 91 (21.9) <0.001  Diabetes mellitus, n (%) 97 (10.9) 4 (1.0) <0.001  Median KDPI/LKDPI (IQR) 69.0 (44.0–93.0) 16.9 (3.0–33.0)  Median HLA mismatches (IQR) 3 (2–4) 3 (2–4) <0.001  Mean cold ischaemia time, hours (SD) 12.1 (5.1) 2.5 (0.8) <0.001 Recipient characteristics  Mean recipient BMI (SD) 25.6 (4.5) 25.2 (4.4) 0.083  Prior kidney transplantation, n (%) 139 (15.6) 22 (5.3) <0.001  Median time on dialysis, months (IQR) 67.0 (39.2–92.0) 10.3 (0.2–29.0) <0.001  Mean raw EPTS (SD) 2.05 (0.66) 1.20 (0.74) <0.001 Recipients’ comorbidities, n (%)  Coronary artery disease 219 (24.6) 48 (11.5) <0.001   With history of MI 71 (8.0) 13 (3.1) 0.001  Diabetes mellitus 141 (15.9) 30 (7.2) <0.001   with end-organ damage 78 (8.8) 19 (4.6) 0.006  Chronic heart failure 70 (7.9) 30 (7.2) 0.737  Peripheral arterial disease 98 (11.0) 26 (6.3) 0.006  Cerebrovascular disease 63 (7.1) 14 (3.4) 0.008  Liver disease 94 (10.6) 33 (7.9) 0.160  Peptic ulcer disease 77 (8.7) 22 (5.3) 0.033  Chronic pulmonary disease 118 (13.3) 44 (10.6) 0.178  Tumour 41 (4.6) 10 (2.4) 0.064  Connective tissue disease 17 (1.9) 14 (3.4) 0.120 Table 1 Donor and recipient characteristics by living and deceased donors Variable Deceased kidney donors Living kidney donors P-value n 889 416 Mean follow-up, years (SD) 6.5 (4.0) 6.5 (4.4) 0.993 Mean recipient age, years (SD) 53.5 (13.7) 43.3 (14.2) <0.001 Mean donor age, years (SD) 54.2 (15.6) 49.8 (11.5) <0.001 Recipient male, n (%) 523 (58.8) 280 (67.3) 0.003 Donor male, n (%) 481 (54.5) 152 (36.5) <0.001 Donor characteristics  Median BMI (IQR) 25.7 (23.6–27.8) 25.1 (22.5–27.8) 0.013  Median creatinine, mg/dL (IQR) 0.9 (0.7–1.2) 0.8 (0.7–0.9) <0.001  Hypertension, n (%) 347 (39.0) 91 (21.9) <0.001  Diabetes mellitus, n (%) 97 (10.9) 4 (1.0) <0.001  Median KDPI/LKDPI (IQR) 69.0 (44.0–93.0) 16.9 (3.0–33.0)  Median HLA mismatches (IQR) 3 (2–4) 3 (2–4) <0.001  Mean cold ischaemia time, hours (SD) 12.1 (5.1) 2.5 (0.8) <0.001 Recipient characteristics  Mean recipient BMI (SD) 25.6 (4.5) 25.2 (4.4) 0.083  Prior kidney transplantation, n (%) 139 (15.6) 22 (5.3) <0.001  Median time on dialysis, months (IQR) 67.0 (39.2–92.0) 10.3 (0.2–29.0) <0.001  Mean raw EPTS (SD) 2.05 (0.66) 1.20 (0.74) <0.001 Recipients’ comorbidities, n (%)  Coronary artery disease 219 (24.6) 48 (11.5) <0.001   With history of MI 71 (8.0) 13 (3.1) 0.001  Diabetes mellitus 141 (15.9) 30 (7.2) <0.001   with end-organ damage 78 (8.8) 19 (4.6) 0.006  Chronic heart failure 70 (7.9) 30 (7.2) 0.737  Peripheral arterial disease 98 (11.0) 26 (6.3) 0.006  Cerebrovascular disease 63 (7.1) 14 (3.4) 0.008  Liver disease 94 (10.6) 33 (7.9) 0.160  Peptic ulcer disease 77 (8.7) 22 (5.3) 0.033  Chronic pulmonary disease 118 (13.3) 44 (10.6) 0.178  Tumour 41 (4.6) 10 (2.4) 0.064  Connective tissue disease 17 (1.9) 14 (3.4) 0.120 Variable Deceased kidney donors Living kidney donors P-value n 889 416 Mean follow-up, years (SD) 6.5 (4.0) 6.5 (4.4) 0.993 Mean recipient age, years (SD) 53.5 (13.7) 43.3 (14.2) <0.001 Mean donor age, years (SD) 54.2 (15.6) 49.8 (11.5) <0.001 Recipient male, n (%) 523 (58.8) 280 (67.3) 0.003 Donor male, n (%) 481 (54.5) 152 (36.5) <0.001 Donor characteristics  Median BMI (IQR) 25.7 (23.6–27.8) 25.1 (22.5–27.8) 0.013  Median creatinine, mg/dL (IQR) 0.9 (0.7–1.2) 0.8 (0.7–0.9) <0.001  Hypertension, n (%) 347 (39.0) 91 (21.9) <0.001  Diabetes mellitus, n (%) 97 (10.9) 4 (1.0) <0.001  Median KDPI/LKDPI (IQR) 69.0 (44.0–93.0) 16.9 (3.0–33.0)  Median HLA mismatches (IQR) 3 (2–4) 3 (2–4) <0.001  Mean cold ischaemia time, hours (SD) 12.1 (5.1) 2.5 (0.8) <0.001 Recipient characteristics  Mean recipient BMI (SD) 25.6 (4.5) 25.2 (4.4) 0.083  Prior kidney transplantation, n (%) 139 (15.6) 22 (5.3) <0.001  Median time on dialysis, months (IQR) 67.0 (39.2–92.0) 10.3 (0.2–29.0) <0.001  Mean raw EPTS (SD) 2.05 (0.66) 1.20 (0.74) <0.001 Recipients’ comorbidities, n (%)  Coronary artery disease 219 (24.6) 48 (11.5) <0.001   With history of MI 71 (8.0) 13 (3.1) 0.001  Diabetes mellitus 141 (15.9) 30 (7.2) <0.001   with end-organ damage 78 (8.8) 19 (4.6) 0.006  Chronic heart failure 70 (7.9) 30 (7.2) 0.737  Peripheral arterial disease 98 (11.0) 26 (6.3) 0.006  Cerebrovascular disease 63 (7.1) 14 (3.4) 0.008  Liver disease 94 (10.6) 33 (7.9) 0.160  Peptic ulcer disease 77 (8.7) 22 (5.3) 0.033  Chronic pulmonary disease 118 (13.3) 44 (10.6) 0.178  Tumour 41 (4.6) 10 (2.4) 0.064  Connective tissue disease 17 (1.9) 14 (3.4) 0.120 The median LKDPI was 16.9. In all, 21.6% of all living donor grafts had a negative LKDPI (better predicted graft survival than every DDK; Figure 1). In contrast, in the US cohort, a median LKDPI of 12.8 (−0.8–27.2) and a slightly higher proportion of grafts (26.5%) with an LKDPI  < 0 were observed by Massie et al. [7]. In our cohort, transplanted DDKs had a large proportion of very high KDPI donor kidneys. This resulted in a considerably higher median KDPI of 69 (IQR 44–93) than for the kidneys recovered for transplantation in the US reference cohort in 2014 (median KDPI by definition 50) (Figure 1). FIGURE 1 View largeDownload slide Distribution of the LKDPI (living donors, n = 416) versus the KDPI (deceased donors, n = 889) of transplanted kidneys in Berlin, 2000–16. FIGURE 1 View largeDownload slide Distribution of the LKDPI (living donors, n = 416) versus the KDPI (deceased donors, n = 889) of transplanted kidneys in Berlin, 2000–16. As expected, there was a significantly better death-censored graft survival after LDK transplantations as compared with DDK transplantations (after 10 years 84 versus 70%; P < 0.001; Figure 2). The best post-transplant creatinine was 1.1 (0.9–1.3) mg/dL versus 1.1 (0.9–1.5) mg/dL (P = 0.017). FIGURE 2 View largeDownload slide Cumulative death-censored graft survival of LDK transplantation versus DDK transplantation in Berlin, 2000–16. FIGURE 2 View largeDownload slide Cumulative death-censored graft survival of LDK transplantation versus DDK transplantation in Berlin, 2000–16. Categorizing LDKs by LKDPI quartiles (LKDPI −44.5–2.9, 3.0–16.8, 17.0–33.0, 33.0–90.0) revealed a significant difference in death-censored graft survival but not in all-cause graft loss (Figure 3). Whereas LDKs of the best LKDPI quartiles (LKDPI −44.5–2.9, 3.0–16.8, 17.0–33.0) tended towards similar graft survival, graft loss occurred significantly more frequently for LDKs of the worst LKDPI quartile (LKDPI 33.0–90.0) for death-censored graft survival (after 10 years 87 versus 73%; P = 0.003) and all-cause graft loss (after 10 years 22 versus 30%; P = 0.041). FIGURE 3 View largeDownload slide Cumulative (A) death-censored graft survival and (B) living with functioning graft as categorized by LKDPI quartiles (LKDPI −45–3, 3–17, 17–33, 33–90) after LDK transplantation in Berlin, 2000–16. FIGURE 3 View largeDownload slide Cumulative (A) death-censored graft survival and (B) living with functioning graft as categorized by LKDPI quartiles (LKDPI −45–3, 3–17, 17–33, 33–90) after LDK transplantation in Berlin, 2000–16. In crude analysis, the KDPI [hazard ratio (HR) 1.15; P < 0.001], the age of the living kidney donor (HR 1.03; P = 0.046) but not the LKDPI (HR 1.11; P = 0.100) were associated with an elevated risk of death-censored graft loss (Table 2). In a multivariate analysis adjusting for age, previous transplantation, time on dialysis and recipient comorbidities (diabetes and coronary artery disease), the KDPI (HR 1.21; P < 0.001), the LKDPI (HR 1.15; P = 0.049) and the age of the living kidney donor (HR 1.03; P = 0.024) were revealed to be significant independent predictors of death-censored graft loss. However, the LKDPI as a predictor for all-cause graft loss neither reached significance in crude nor in adjusted analysis (HR 1.08; P = 0.136 and HR 1.06; P = 0.331). The KDPI’s adjusted hazard of all-cause graft loss was 1.15 per 10 increment (P < 0.001), thus slightly higher than the hazard reported by Massie et al. [7] in the US cohort (HR 1.10 per 10 increment; P < 0.001). Table 2 Univariate and multivariate Cox regression analysis of the LKDPI, KDPI and living donor age associated with an elevated risk of death-censored graft loss and all-cause graft loss in LDK and DDK transplant recipients Variable All-cause graft loss Death-censored graft loss HR (95% CI) P-value HR (95% CI) P-value Monovariate analysis  LKDPI per 10 increment (n = 416) 1.08 (0.98–1.20) 0.136 1.11 (0.98–1.27) 0.100  Living donor age (n = 416) 1.02 (1.00–1.05) 0.029 1.03 (1.00–1.06) 0.046  KDPI per 10 increment (n = 889) 1.17 (1.12–1.22) <0.001 1.15 (1.08–1.22) <0.001 Multivariate analysis  LKDPI per 10 incrementa (n = 416) 1.06 (0.95–1.18) 0.331 1.15 (1.00–1.31) 0.049  Living donor agea (n = 416) 1.02 (1.00–1.04) 0.072 1.03 (1.00–1.06) 0.024  KDPI per 10 incrementa (n = 889) 1.15 (1.09–1.21) <0.001 1.21 (1.13–1.30) <0.001 Variable All-cause graft loss Death-censored graft loss HR (95% CI) P-value HR (95% CI) P-value Monovariate analysis  LKDPI per 10 increment (n = 416) 1.08 (0.98–1.20) 0.136 1.11 (0.98–1.27) 0.100  Living donor age (n = 416) 1.02 (1.00–1.05) 0.029 1.03 (1.00–1.06) 0.046  KDPI per 10 increment (n = 889) 1.17 (1.12–1.22) <0.001 1.15 (1.08–1.22) <0.001 Multivariate analysis  LKDPI per 10 incrementa (n = 416) 1.06 (0.95–1.18) 0.331 1.15 (1.00–1.31) 0.049  Living donor agea (n = 416) 1.02 (1.00–1.04) 0.072 1.03 (1.00–1.06) 0.024  KDPI per 10 incrementa (n = 889) 1.15 (1.09–1.21) <0.001 1.21 (1.13–1.30) <0.001 a Adjusted for recipient age, time on dialysis, prior kidney transplantation, diabetes and coronary artery disease. Table 2 Univariate and multivariate Cox regression analysis of the LKDPI, KDPI and living donor age associated with an elevated risk of death-censored graft loss and all-cause graft loss in LDK and DDK transplant recipients Variable All-cause graft loss Death-censored graft loss HR (95% CI) P-value HR (95% CI) P-value Monovariate analysis  LKDPI per 10 increment (n = 416) 1.08 (0.98–1.20) 0.136 1.11 (0.98–1.27) 0.100  Living donor age (n = 416) 1.02 (1.00–1.05) 0.029 1.03 (1.00–1.06) 0.046  KDPI per 10 increment (n = 889) 1.17 (1.12–1.22) <0.001 1.15 (1.08–1.22) <0.001 Multivariate analysis  LKDPI per 10 incrementa (n = 416) 1.06 (0.95–1.18) 0.331 1.15 (1.00–1.31) 0.049  Living donor agea (n = 416) 1.02 (1.00–1.04) 0.072 1.03 (1.00–1.06) 0.024  KDPI per 10 incrementa (n = 889) 1.15 (1.09–1.21) <0.001 1.21 (1.13–1.30) <0.001 Variable All-cause graft loss Death-censored graft loss HR (95% CI) P-value HR (95% CI) P-value Monovariate analysis  LKDPI per 10 increment (n = 416) 1.08 (0.98–1.20) 0.136 1.11 (0.98–1.27) 0.100  Living donor age (n = 416) 1.02 (1.00–1.05) 0.029 1.03 (1.00–1.06) 0.046  KDPI per 10 increment (n = 889) 1.17 (1.12–1.22) <0.001 1.15 (1.08–1.22) <0.001 Multivariate analysis  LKDPI per 10 incrementa (n = 416) 1.06 (0.95–1.18) 0.331 1.15 (1.00–1.31) 0.049  Living donor agea (n = 416) 1.02 (1.00–1.04) 0.072 1.03 (1.00–1.06) 0.024  KDPI per 10 incrementa (n = 889) 1.15 (1.09–1.21) <0.001 1.21 (1.13–1.30) <0.001 a Adjusted for recipient age, time on dialysis, prior kidney transplantation, diabetes and coronary artery disease. Model discrimination was superior for the KDPI compared with the LKDPI. The predictive strength for death-censored graft survival or living with a functioning graft investigated by the AUC was 0.55 and 0.53 for the LKDPI and 0.66 and 0.65 for the KDPI , respectively (Figure 4). As there was limited predictive strength investigating the AUC of a continuous LKDPI, discrimination was examined in a dichotomous way (LKDPI  <33 versus LKDPI  ≥33) and hence a better discrimination was obtained (AUC of 0.60 for death-censored graft survival and AUC of 0.54 for living with a functioning graft). Furthermore, these results were qualitatively reproduced in a subgroup of 323 LDK recipients who had complete data for calculation of the LKDPI (no missing values except ethnicity, which is not routinely filed in Germany). Although limited due to the smaller sample size and shorter follow-up, the discriminative ability of the LKDPI was comparable and the C-statistic resulted in an AUC of 0.58 for death-censored graft loss and 0.52 for all-cause graft loss up to 8 years post-transplant in this subgroup. By calculating the discrimination of a refit model, Massie et al. [7] reported a C-statistic of 0.59 for the LKDPI and 0.58 for the KDPI. FIGURE 4 View largeDownload slide Time-to-event ROC curves for the LKDPI (n = 416) and KDPI (n = 889) for (A) death-censored graft survival and (B) living with functioning graft. FIGURE 4 View largeDownload slide Time-to-event ROC curves for the LKDPI (n = 416) and KDPI (n = 889) for (A) death-censored graft survival and (B) living with functioning graft. In subgroups with corresponding LKDPI/KDPI (0–20, 20–40), we compared the incidence of death-censored graft loss and all-cause graft loss of LDKs versus DDKs. Results were comparable and did not differ significantly [death-censored graft survival P = 0.704 and P = 0.711 (Figure 5); all-cause graft loss P = 0.231 and P = 0.949]. FIGURE 5 View largeDownload slide Recipients' cumulative death-censored graft survival after LDK transplantation as categorized by the LKDPI versus DDK transplantation as categorized by the KDPI: (A) LKDPI/KDPI 0–20, (B) 20–40. FIGURE 5 View largeDownload slide Recipients' cumulative death-censored graft survival after LDK transplantation as categorized by the LKDPI versus DDK transplantation as categorized by the KDPI: (A) LKDPI/KDPI 0–20, (B) 20–40. Comparing the imputed eGFR of the complete cohort showed a significantly better eGFR for LDK recipients than for DDK recipients at all time points (Figure 6). Nevertheless, the comparison of the eGFR in subgroups with corresponding LKDPI/KDPI revealed no significant differences in the imputed eGFR of 49 versus 56 mL/min/1.73 m2 (P = 0.187, LKDPI/KDPI 0–20) and versus 44 mL/min/1.73 m2 (P = 0.707, LKDPI/KDPI 20–40) at 8 years post-transplant. Interestingly, the 10-year survival of LDK and DDK recipients also proved to be comparable [LKDPI/KDPI 0–20: 86% and 81%; P = 0.331 and LKDPI/KDPI 20–40: 89% (both); P = 0.449]. LDK transplantations did not contribute an independent benefit to improve death-censored graft survival in a multivariate analysis adjusted for recipient age, time on dialysis, prior kidney transplantation, diabetes and coronary artery disease (LKDPI/KDPI 0–20: HR 1.40, P = 0.480; 20–40: HR 1.79, P = 0.325). FIGURE 6 View largeDownload slide Median eGFR using imputation for values after graft loss (patients with graft loss: GFR = 0 mL/min/1.73 m2). eGFR was calculated using the CKD-EPI formula after LDK transplantation versus DDK transplantation: (A) all recipients, (B) LKDPI/KDPI 0–20, (C) LKDPI/KDPI 20–40. FIGURE 6 View largeDownload slide Median eGFR using imputation for values after graft loss (patients with graft loss: GFR = 0 mL/min/1.73 m2). eGFR was calculated using the CKD-EPI formula after LDK transplantation versus DDK transplantation: (A) all recipients, (B) LKDPI/KDPI 0–20, (C) LKDPI/KDPI 20–40. The aggregate cumulative incidence of all-cause graft loss after 8 years was roughly 20% for American LDK recipients, who received an LDK with an LKDPI  < 0, and roughly 40% for recipients who received an LDK with an LKDPI >40 [20]. At our transplant centre, all-cause graft loss after 8 years occurred less frequently (12% and 24%, respectively). DISCUSSION For the first time, we provide external validation of the recently proposed LKDPI in a non-American cohort. Although the choice of LDK transplantation is highly driven by the living donor’s risk of end-stage renal disease, it may be valuable to assess the donor kidney quality in order to compare LDKs with each other and with DDKs [21]. LDK quality as defined by the LKDPI was slightly better in the US cohort analysed by Massie et al. [7] than in this European cohort (LKDPI 12.8 versus 16.9). The US allocation system implemented the KDPI for DDKs in late 2014. Its purpose is to match the longevity of the donor kidney with the estimated survival of the recipient as defined by the EPTS score [8]. In addition, the dichotomous prior allocation routine was thought to complicate the utilization of marginal donor grafts. The KDPI is also a useful tool to compare DDK quality recovered in different countries. A high proportion of very high KDPI kidneys allocated within the Eurotransplant system were transplanted at our centre, resulting in a median KDPI considerably higher than in the US cohort. In 2013, less than one-tenth (9%) of DDKs recovered for transplantation by the US allocation system had a KDPI >85, whereas 35% of all DDK recipients of our cohort received such a graft. Moreover, these very high KDPI kidneys have a high discard rate (55%) in the USA [22]. Since the KDPI is not implemented within the Eurotransplant system, there are few data about KDPI-assessed DDK quality transplanted within other Eurotransplant member countries. It is likely that differences may be expected as donation rates between Eurotransplant countries and policies between transplant centres show divergences. A prospective single-centre study at the Antwerp University Hospital in Belgium calculated a considerably lower median KDPI [47 (IQR 33–58)] with an inferior proportion of DDKs with a KDPI >85 (1.3%) compared with our DDK cohort [23]. However, a combined retro- and prospective multicentre study from Italy included 248 standard criteria donors [mean KDPI 35 (SD 19.2)] and 442 marginal donors (≥65 years or renal dysfunction) and thus received a higher KDPI for the whole cohort [66.6 (SD 29.6)] with a high proportion of high KDPI kidneys (30% KDPI  ≥91) [10]. Of note, all consecutive DDK transplantations in our centre were included and the proportion of donors ≥ 65 years was 30%. Finally, a retrospective multicentre study from The Netherlands revealed a median KDRI of 1.21, comparable with a median KDRI of 1.24 in the USA in 2012 [24]. Encouraged by promising follow-up data, our transplant centre pursues a guideline-directed selection of the LDK and DDK donor aimed at a good functional match. Moreover, especially in patients who received marginal organs, the transplant outcome depends on close post-transplant monitoring. As expected, LDK recipients had a significantly better graft survival than DDK recipients in this cohort despite an overlap between LDK and DDK quality as shown by Massie et al. [7] and confirmed in this cohort. There are relevant differences comparing death-censored graft survival in the USA and in our cohort. For instance, LDKs transplanted in 2003 in the USA showed a 10-year death-censored graft survival of 79 versus 84% in our cohort [22]. Comparably, although a higher proportion of very high KDPI kidneys was transplanted at our centre, US recipients of DDKs when transplanted in 2003 showed a similar death-censored graft survival compared with DDK recipients at our centre (after 10 years, 72 versus 70%) [22]. Whether different post-transplant care and insurance policies (especially as parts of US Medicare ensure immunosuppressive drug coverage only up to 36 months post-transplant) could explain these survival disparities to some extent is a question beyond the scope of this study [13, 25]. Previous studies reported higher mortality as well as worse graft survival in recipients showing multiple comorbidities [26–28]. In this regard, confounding recipient characteristics should be considered to assess the LKDPI’s performance. Although LDK recipients at our centre were significantly younger and showed less prevalence of comorbidities than DDK recipients, a relevant proportion had important comorbidities such as coronary artery disease or diabetes mellitus. Indeed, an adjusted analysis for recipients’ comorbidities and allocation factors confirmed the LKDPI as a significant and independent risk factor of death-censored graft survival in our cohort of 416 LDK recipients (Table 2). The LKDPI demonstrated only moderate discrimination in this cohort (C-statistic of 0.55 for death-censored graft survival). Massie et al. [7] hypothesized that the range of quality is narrower for LDK than for DDK donors, especially as DDK donors show a broader spectrum of comorbidities. Notably, the IQR as an indicator of the range of quality was also narrower for LDKs of our cohort (LDK 30 versus DDK 49). Schold et al. [20] addressed the limited predictive power of the LKDPI and advised cautious application to individual decision-making. Probably the LKDPI’s low discrimination may simply emphasize the unneglectable effect of recipient factors on graft and patient survival [7]. Further investigation in other cohorts may be necessary to better investigate the LKDPI’s ability to guide the individual patient. In our cohort, the specificity and sensitivity of the LKDPI tended to be higher for high LKDPI kidneys (Figure 4), thus suggesting further analysis to examine the LKDPI as a dichotomous variable. As our data indicate a survival disadvantage in particular for LKDPI kidneys in the worst LKDPI quartile, we analysed the predictive strength of LKDPI  ≥33 versus LKDPI <33. Indeed, a considerably higher predictive power results for death-censored graft survival. In a dichotomous view, LKDPI  ≥33 is also a strong and significant risk factor of death-censored graft survival (adjusted HR 3.01; P = 0.001). Regarding DDK transplantation in the USA, there is a dichotomous use of the KDPI in several cases. For example, high KDPI kidneys (KDPI >85) are designated to patients who give a candidate consent and the best KDPI kidneys are assigned to recipients with an EPTS  ≤20. However, to evaluate how this practice could apply to the LKDPI exceeds the scope of this study. Our findings are in line with prior reported data of an increasing C-statistic in high KDRI transplants [29]. The discrimination of the KDPI derived from DDKs transplanted at our centre was similar to previously published data [8]. Massie et al. [7] reported higher donor age among living donors >50 years of age as a significant predictor of reduced all-cause graft loss. Living donor age was also investigated by Noppakun et al. [5]. They reported an association between increasing donor age and worse outcome that was observed particularly after 4 years post-transplant and modified by recipient age. Furthermore, several other studies reported donor age as a significant predictor of reduced graft survival when used as a dichotomous parameter [6, 30, 31]. We investigated older living donor age as a continuous predictor of death-censored graft loss in an adjusted analysis, confirming a significant predictive value with an HR of 1.03 per year. Thus our data emphasize the great influence of increasing age of the living kidney donor regardless of the age percentile. Additionally, the hazard of increasing age of the living kidney donor increases when calculated in the subgroup of donors >50 years of age (HR 1.17; P < 0.001). This is of particular interest as since, on the one hand, there is a very high proportion of living donors >50 years of age in our cohort (50%) and, on the other hand, the age of living kidney donors at transplantation generally increases [3]. Massie et al. [7] developed the LKDPI for comparison of LDKs with DDKs. As intended by the LKDPI algorithm, our analysis of LDKs also showed similar graft survival and post-transplant eGFR compared with DDKs in subgroups of corresponding LKDPI/KDPI. Further analysis revealed comparable patient survival and no significant benefit of an LDK donation according to an adjusted multivariate analysis taking relevant recipient factors into account. Since the LKDPI is able to significantly stratify the risk of death-censored graft loss, these findings indicate comparability of the LKDPI and KDPI in a certain range of kidney quality. By performing an observational study, we could only rely on data recorded at our centre. Some data were missing in a minority of living donors. For example, since ethnic information is not routinely recorded in the Eurotransplant system, for calculation of the LKDPI, all patients were classified as Caucasian according to the population structure in Germany. While the original study excluded previously transplanted patients, we included these patients and adjusted for this characteristic in a multivariate Cox analysis. These results provide external validation of the LKDPI as a tool to assess LDK quality and prognosis of graft survival. The LKDPI may be valuable to compare LDKs with each other and with DDKs, although the moderate discriminative ability indicates limited clinical use for guidance of the individual patient. Comparing the outcome of this cohort with US data, a worse death-censored graft survival for LDK recipients in the USA is shown. In addition, comparing corresponding subgroups of LDKs (LKDPI  <0 or LKDPI >40) revealed a higher all-cause graft loss at 8 years post-transplant in the USA. AUTHORS’ CONTRIBUTIONS O.S., G.R., A.K., A.S. and M.D. were responsible for the conception and design or analysis and interpretation of data, or both. G.R., O.S., F.H., D.K., L.L. and K.B. were responsible for drafting the article or revising it. G.R., O.S., K.B., F.H., D.K., L.L. and M.D. provided intellectual content of critical importance to the work described. G.R., F.H., D.K., L.L, A.K., A.S., M.D., K.B. and O.S. approved the final version to be published. CONFLICT OF INTEREST STATEMENT None declared. The results presented in this article have not been published previously in whole or part, except in abstract format. REFERENCES 1 Cozzi E , Biancone L , Lopez-Fraga M et al. Long-term outcome of living kidney donation: position paper of the European Committee on Organ Transplantation, Council of Europe . Transplantation 2016 ; 100 : 270 – 271 Google Scholar CrossRef Search ADS PubMed 2 Hart A , Smith JM , Skeans MA et al. Kidney . Am J Transplant 2016 ; 16 : 11 – 46 Google Scholar CrossRef Search ADS PubMed 3 Hart A , Smith JM , Skeans MA et al. OPTN/SRTR 2015 annual data report: kidney . Am J Transplant 2017 ; 17 : 21 – 116 Google Scholar CrossRef Search ADS PubMed 4 Araujo AM , Santos F , Guimaraes J et al. Living-donor kidney transplantation: predictive factors and impact on post-transplant outcome . Transplant Proc 2015 ; 47 : 938 – 941 Google Scholar CrossRef Search ADS PubMed 5 Noppakun K , Cosio FG , Dean PG et al. Living donor age and kidney transplant outcomes . Am J Transplant 2011 ; 11 : 1279 – 1286 Google Scholar CrossRef Search ADS PubMed 6 Fuggle SV , Allen JE , Johnson RJ et al. Factors affecting graft and patient survival after live donor kidney transplantation in the UK . Transplantation 2010 ; 89 : 694 – 701 Google Scholar CrossRef Search ADS PubMed 7 Massie AB , Leanza J , Fahmy LM et al. A risk index for living donor kidney transplantation . Am J Transplant 2016 ; 16 : 2077 – 2084 Google Scholar CrossRef Search ADS PubMed 8 Lee AP , Abramowicz D. Is the Kidney Donor Risk Index a step forward in the assessment of deceased donor kidney quality? Nephrol Dial Transplant 2015 ; 30 : 1285 – 1290 Google Scholar CrossRef Search ADS PubMed 9 Bae S , Massie AB , Luo X et al. Changes in discard rate after the introduction of the Kidney Donor Profile Index (KDPI) . Am J Transplant 2016 ; 16 : 2202 – 2207 Google Scholar CrossRef Search ADS PubMed 10 Gandolfini I , Buzio C , Zanelli P et al. The Kidney Donor Profile Index (KDPI) of marginal donors allocated by standardized pretransplant donor biopsy assessment: distribution and association with graft outcomes . Am J Transplant 2014 ; 14 : 2515 – 2525 Google Scholar CrossRef Search ADS PubMed 11 Tanriover B , Mohan S , Cohen DJ et al. Kidneys at higher risk of discard: expanding the role of dual kidney transplantation . Am J Transplant 2014 ; 14 : 404 – 415 Google Scholar CrossRef Search ADS PubMed 12 Gupta A , Chen G , Kaplan B. KDPI and donor selection . Am J Transplant 2014 ; 14 : 2444 – 2445 Google Scholar CrossRef Search ADS PubMed 13 Ojo AO , Morales JM , Gonzalez-Molina M et al. Comparison of the long-term outcomes of kidney transplantation: USA versus Spain . Nephrol Dial Transplant 2013 ; 28 : 213 – 220 Google Scholar CrossRef Search ADS PubMed 14 Gore JL , Danovitch GM , Litwin MS et al. Disparities in the utilization of live donor renal transplantation . Am J Transplant 2009 ; 9 : 1124 – 1133 Google Scholar CrossRef Search ADS PubMed 15 Levey AS , Stevens LA , Schmid CH et al. A new equation to estimate glomerular filtration rate . Ann Intern Med 2009 ; 150 : 604 – 612 Google Scholar CrossRef Search ADS PubMed 16 Rao PS , Schaubel DE , Guidinger MK et al. A comprehensive risk quantification score for deceased donor kidneys: the kidney donor risk index . Transplantation 2009 ; 88 : 231 – 236 Google Scholar CrossRef Search ADS PubMed 17 OPTN/UNOS . A Guide to Calculating and Interpretating the Kidney Donor Profile Index (KDPI). Richmond, VA: Organ Procurement and Transplantation Network, 2016. Available at: https://optn.transplant.hrsa.gov/media/1512/guide_to_calculating_interpreting_kdpi.pdf (25 April 2018, date last accessed) 18 OPTN . A Guide to Calculating and Interpretating the Estimated Post-Transplant Survival (EPTS) Score Used in the Kidney Allocation System (KAS). Richmond, VA: Organ Procurement and Transplantation Network, 2014. Available at: https://optn.transplant.hrsa.gov/media/1511/guide_to_calculating_interpreting_epts.pdf (25 April 2018, date last accessed) 19 Heagerty PJ , Lumley T , Pepe MS. Time-dependent ROC curves for censored survival data and a diagnostic marker . Biometrics 2000 ; 56 : 337 – 344 Google Scholar CrossRef Search ADS PubMed 20 Schold JD , Kaplan B. Living kidney donor profile index: utility and limitations . Am J Transplant 2016 ; 16 : 1951 – 1952 Google Scholar CrossRef Search ADS PubMed 21 Grams ME , Garg AX , Lentine KL. Kidney-failure risk projection for the living kidney-donor candidate . N Engl J Med 2016 ; 374 : 2094 – 2095 Google Scholar CrossRef Search ADS PubMed 22 Matas AJ , Smith JM , Skeans MA. OPTN/SRTR 2013 annual data report: kidney . Am J Transplant 2015 ; 15 (Suppl 2): 1 – 34 Google Scholar CrossRef Search ADS PubMed 23 Philipse E , Lee APK , Bracke B et al. Does Kidney Donor Risk Index implementation lead to the transplantation of more and higher-quality donor kidneys? Nephrol Dial Transplant 2017 ; 32 : 1934 – 1938 Google Scholar CrossRef Search ADS PubMed 24 Peters-Sengers H , Heemskerk MBA , Geskus RB et al. Validation of the prognostic kidney donor risk index scoring system of deceased donors for renal transplantation in the Netherlands . Transplantation 2018 ; 102 : 162 – 170 Google Scholar CrossRef Search ADS PubMed 25 Tanriover B , Stone PW , Mohan S et al. Future of Medicare immunosuppressive drug coverage for kidney transplant recipients in the United States . Clin J Am Soc Nephrol 2013 ; 8 : 1258 – 1266 Google Scholar CrossRef Search ADS PubMed 26 Jassal SV , Schaubel DE , Fenton SS. Baseline comorbidity in kidney transplant recipients: a comparison of comorbidity indices . Am J Kidney Dis 2005 ; 46 : 136 – 142 Google Scholar CrossRef Search ADS PubMed 27 Wu C , Evans I , Joseph R et al. Comorbid conditions in kidney transplantation: association with graft and patient survival . J Am Soc Nephrol 2005 ; 16 : 3437 – 3444 Google Scholar CrossRef Search ADS PubMed 28 Halleck F , Khadzhynov D , Lehner L et al. Impact of pre-existing comorbidities on long-term patient and graft survival in kidney transplant recipients . Am J Transplant 2016 ; 16 (Suppl 3): 489 – 489 29 Akkina SK , Asrani SK , Peng Y et al. Development of organ-specific donor risk indices . Liver Transpl 2012 ; 18 : 395 – 404 Google Scholar CrossRef Search ADS PubMed 30 Issa N , Stephany B , Fatica R et al. Donor factors influencing graft outcomes in live donor kidney transplantation . Transplantation 2007 ; 83 : 593 – 599 Google Scholar CrossRef Search ADS PubMed 31 Young A , Kim SJ , Speechley MR et al. Accepting kidneys from older living donors: impact on transplant recipient outcomes . Am J Transplant 2011 ; 11 : 743 – 750 Google Scholar CrossRef Search ADS PubMed © The Author(s) 2018. Published by Oxford University Press on behalf of ERA-EDTA. All rights reserved. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)

Journal

Nephrology Dialysis TransplantationOxford University Press

Published: May 9, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off