Abstract
The effect of baseline histology and individual histologic lesions at the time of transplantation on long-term graft survival has been evaluated using different scoring systems, but the predictive capacity of these systems has not been adequately validated. All kidney recipients transplanted in a single institution between 1991 and 2009 who underwent a baseline kidney allograft biopsy at transplantation were included in this prospective study (N=548). All baseline biopsies were rescored according to the updated Banff classification, and the relationship between the individual histologic lesions and donor demographics was assessed using hierarchical clustering and principal component analysis. Survival analysis was performed using Cox proportional hazards analysis and log-rank testing. Mean follow-up time was 6.7 years after transplantation. Interstitial fibrosis, tubular atrophy, and glomerulosclerosis associated significantly with death-censored graft survival, whereas arteriolar hyalinosis and vascular intimal thickening did not. Notably, donor age correlated significantly with interstitial fibrosis, tubular atrophy, and glomerulosclerosis and associated independently with graft survival. On the basis of these findings, a novel scoring system for prediction of 5-year graft survival was constructed by logistic regression analysis. Although the predictive performance of previously published histologic scoring systems was insufficient to guide kidney allocation in our cohort (receiver operating characteristic area under the curve ≤0.62 for each system), the new system based on histologic data and donor age was satisfactory for prediction of allograft loss (receiver operating characteristic area under the curve = 0.81) and may be valuable in the assessment of kidney quality before transplantation.
The demand for kidney grafts exceeds the supply of available organs. As a result, there is an increasing use of older and extended criteria donor kidneys. This is not without consequences. The risk of graft failure after an extended criteria donor kidney transplant is 70% higher than after transplantation with a standard criteria donor.1 The intrinsic quality of the kidney at transplantation becomes increasingly important for the post-transplant histologic evolution and long-term graft survival.2–4
To evaluate the intrinsic donor organ quality, baseline biopsies are increasingly used by many transplant centers. Pretransplantation kidney biopsies are mostly evaluated in older donors and are a significant reason for discarding potential grafts. In 2001, 47% of kidneys in the United States were discarded based on biopsy results. The odds of discard increase progressively with increasing degrees of glomerulosclerosis and many transplant centers reject kidneys with extensive glomerulosclerosis (e.g., >20%).5 Another approach has been to consider baseline histology when deciding between single and dual kidney transplantation, as proposed by Remuzzi et al,6,7 who showed that the risk of graft failure in patients aged >60 years was 3.68 times greater when transplanted kidneys were not histologically assessed and these results were not used for kidney allocation. However, both kidney discard and dual kidney allocation lead to a decrease in the absolute number of effective kidney transplants, and thus to an increase in waiting time for deceased donor kidney transplantation. Detailed study of implantation biopsy histology, and its prognostic value, is therefore necessary to make best use of the available kidneys for transplantation.
Several centers have studied the value of baseline biopsies of kidneys procured for kidney transplantation. Several studies correlated baseline biopsy data with post-transplant graft function, but not with graft survival,8–11 whereas others showed a significant association between baseline histology and graft survival.6,12–17 On the basis of this association between baseline histology and post-transplant graft survival, the following histologic scoring schemes have been proposed for prediction of graft survival based on donor biopsy histology: the Remuzzi score,6 the donor chronic damage score,14 the Maryland aggregate pathology index (MAPI),15 and the total chronic Banff score.16 The score proposed by Remuzzi et al.6 was associated with graft survival, and using this histologic score for decision of dual kidney transplantation was beneficial for graft outcome. However, this score was not validated in an independent population and the predictive capacity of this specific score for graft outcome and for kidney discard remains unclear. The donor chronic damage score was also significantly associated with graft survival, but the predictive performance of this score has not been assessed.14 The MAPI needs potentially cumbersome morphometric analysis of the arterial wall-to-lumen ratio, which is not readily available in many centers.15 In addition, the construction of this scoring system is subject to selection bias, because the implantation biopsy histologic features were used for kidney acceptance and allocation.15 A previous European study that explicitly excluded this type of selection bias evaluated the predictive capacity of the total chronic Banff score in implantation histology for graft survival, but only in a small subgroup of older donors after cardiac death.16 On the basis of data obtained in 52 donor biopsies, these authors concluded that a total chronic Banff score ≥4 was associated with unacceptable graft survival. Replication of these findings in a larger cohort, and in donors after brain death, has not been performed.
Thus, both the assesment of preimplantation biopsies for allocation purposes and the cut off values or scores that could be used for discarding a kidney are currently not known. In this study, we evaluated the effect of the histologic appearance of implantation biopsies on long-term renal allograft outcome, and assessed the histology of implantation biopsies as a prognostic clinical tool.
Results
Population Characteristics
The study group consisted of 548 kidney recipients who had a baseline biopsy performed at the time of transplantation. Six biopsies were of insufficient quality, leaving 542 baseline biopsies for analysis. Table 1 summarized the recipient and donor characteristics. The mean recipient age was 51.9±13.4 years (range, 15–77). Most recipients received their first allograft (88%). Mean donor age was 43.2±15.6 years (range, 8–74). Brain death was caused by stroke in 40.5% of the donors. A history of hypertension was reported in 15.7% of donors (n=86) and 1.3% (n=7) had reported diabetes. Delayed graft function occurred in 21% of recipients (n=119). Ninety-five percent of patients reached at least 1-year graft survival. The 2-, 5-, and 10-year death-censored graft survival rates were 93.5%, 89.8%, and 81.9%. At the time of our data extraction, the mean post-transplant time of functioning grafts was 5.5±3.2 years (10.7±1.7 in the historic cohort and 4.3±1.8 in the validation cohort). The demographics of the historic cohort, in which baseline biopsies were performed not systematically but only at the request of the transplant surgeon, were similar to the demographics of the validation cohort (Table 1).
Table 1.
Demographic and clinical characteristics of transplant donors and recipients (N=548)
| Characteristic | All (N=548) | Historic Cohort (n=181) | Validation Cohort (n=367) | P |
|---|---|---|---|---|
| Recipients | ||||
| Age (yr) | 51.9±13.4 | 51.4±13.5 | 52.2±13.4 | <0.001 |
| Men | 327 (60.3) | 117 (64) | 210 (57) | 0.12 |
| Repeat transplantation | 66 (12) | 17 (9.0) | 49 (13.4) | 0.21 |
| Donors | ||||
| Age (yr) | 43.16±15.65 | 40.9±15 | 44.2±14 | <0.001 |
| Men | 301 (53.9) | 104 (57.4) | 197 (53.7) | 0.41 |
| Deceased donor | 544 (99.2) | 180 (99.4) | 363 (98.9) | 1.00 |
| Brain death/cardiac death | 222 (40.5)/326 (59.5) | 74 (40.8)/107 (59.2) | 148 (40.3)/219 (59.9) | 0.92 |
| Stroke as reason for death | 301 (40.5) | 74 (40.8) | 177 (48.2) | 0.92 |
| History of hypertension | 86 (15.7) | 24 (13.2) | 62 (16.9) | 0.32 |
| History of diabetes mellitus | 7 (1.3) | 1 (0.5) | 6 (1.6) | 0.43 |
| History of smoking | 127 (23.1) | 32 (17.6) | 95 (25.8) | 0.04 |
Data are presented as the mean ± SD or n (%). P values compare the historic versus the validation cohort, and were calculated using the chi-squared test or two-way ANOVA.
Association between Clinical Donor Characteristics, Baseline Histology, and Graft Survival
In univariate Cox proportional hazards analysis of the donor demographics and death-censored graft survival, only older donor age (hazard ratio [HR], 1.72 per year; 95% confidence interval [95% CI], 1.2 to 2.46; P=0.003) was a significant predictor of long-term graft loss (Figure 1 and Table 2). The other donor factors (hypertension, smoking, stroke as reason of donor death, cold ischemia time, and diabetes mellitus) were not associated with post-transplant graft outcome. Table 3 shows the distribution of the histopathologic lesions in the biopsies of sufficient quality and diagnosis is shown.
Figure 1.
Association of death-censored graft survival and donor age, by Kaplan–Meier survival analysis. The P value is calculated with the log-rank test. TX, transplantation.
Table 2.
Univariate and multivariate Cox proportional hazards analysis of death-censored graft survival based on baseline histology and donor demographic variables (n=542)
| Variables | Univariate Analysis | Multivariate Analysis | ||
|---|---|---|---|---|
| Hazard Ratio (95% Confidence Interval) | P | Hazard Ratio (95% Confidence Interval) | P | |
| Donor demographics | ||||
| Donor age (<40 versus 40–60 versus >60 yr) | 1.72 (1.2 to 2.46) | 0.003 | 1.90 (1.24 to 2.9) | 0.003 |
| Stroke as reason of donor death | 1.38 (0.83 to 2.29) | 0.60 | – | – |
| Donor history of hypertension | 1.18 (0.61 to 2.29) | 0.62 | – | – |
| Donor diabetes mellitus | 1.69 (0.23 to 12.2) | 0.60 | – | – |
| Donor history of smoking | 0.90 (0.49 to 1.69) | 0.75 | – | – |
| Baseline biopsy histology | ||||
| Tubular atrophy (present versus absent) | 1.60 (0.99 to 2.58) | 0.05 | 0.98 (0.59 to 1.76) | 0.90 |
| Arteriolar hyalinosis (present versus absent) | 1.230 (0.71 to 2.38) | 0.40 | – | – |
| Interstitial fibrosis (present versus absent) | 1.85 (1.32 to 2.57) | 0.0003 | 1.51 (1.02 to 2.26) | 0.03 |
| Vascular intimal thickening (present versus absent) | 1.08 (0.26 to 4.43) | 0.92 | – | – |
| Glomerulosclerosis (≥10% versus <10%) | 1.98 (1.21 to 3.26) | 0.01 | 1.32 (0.73 to 2.39) | 0.35 |
Table 3.
Distribution of the histologic lesions in 542 implantation biopsies
| Histological Lesions | Banff Grade | |||
|---|---|---|---|---|
| 0 | 1 | 2 | 3 | |
| Arteriolar hyalinosisa | 446 (82.7) | 76 (14.1) | 15 (2.7) | 2 (0.3) |
| Interstitial fibrosis | 407 (75) | 99 (18) | 26 (4.7) | 10 (1.8 |
| Tubular atrophy | 287 (52.9) | 240 (44.2) | 12 (2.2) | 3 (0.5) |
| Vascular intimal thickening | 519 (97.5) | 21 (3.9) | 2 (0.3) | 0 (0.0) |
| Glomerulosclerosisa,b | 420 (78.2) | 80 (14.9) | 37 (6.9) | – |
Data are presented as n (%). The mean number of glomeruli was per biopsy was 21.3±15.1 (range, 1–89).
Not all biopsies had sufficient numbers of glomeruli or arterioles to evaluate this lesion.
Grade 1, glomerulosclerosis of <10% of glomeruli; grade 2, glomerulosclerosis in 10%–25% of glomeruli; grade 3, glomerulosclerosis of >25% of glomeruli.
The mean number of glomeruli measured in the biopsies was 21.3±15.2 (range, 1–89). Twenty-two percent of patients (n=117) had >10% glomerulosclerosis. Our results showed that 19.1% of biopsies (n=105) showed interstitial fibrosis/tubular atrophy and 17.3% of biopsies had arteriolar hyalinosis (n=93). Only 23 biopsies (4.3%) showed vascular intimal thickening. Univariate analysis showed that interstitial fibrosis (grade > 0 versus = 0: HR, 1.98; 95% CI, 1.32 to 2.57; P=0.0003), tubular atrophy (grade > 0 versus = 0: HR, 1.60; 95% CI, 0.99 to 2.58; P=0.05), and glomerulosclerosis (>10% versus <10%: HR, 1.98; 95% CI, 1.21 to 3.26; P=0.01) associated with death-censored graft survival, whereas arteriolar hyalinosis (grade > 0 versus = 0: HR, 1.30; 95% CI, 0.71 to 2.38; P=0.4) and vascular intimal thickening (grade > 0 versus = 0: HR, 1.08; 95% CI, 0.26 to 4.43; P=0.92) did not (Figure 2 and Table 2). In multivariate analysis, only interstitial fibrosis (adjusted HR, 1.51; 95% CI, 1.02 to 2.26; P=0.03) was independently associated with graft survival, adjusted for donor age (Table 2).
Figure 2.
Association of death-censored graft survival and selected individual histologic lesions in 542 baseline biopsies, by Kaplan–Meier survival analysis. P values are calculated with the log-rank test. TX, transplantation.
Clinical Determinants of Baseline Histology
The correlation between donor demographic variables and baseline histology is provided in Table 4. Donor age was highly significantly correlated with interstitial fibrosis, tubular atrophy, and glomerulosclerosis (all P<0.001), but less with arteriolar hyalinosis and vascular intimal thickening. Vascular intimal thickening was highly significantly associated with donor history of diabetes, and arteriolar hyalinosis was most significantly associated with stroke and donor history of hypertension (Table 4).
Table 4.
Spearman correlation between the histologic lesions in implantation biopsies and clinical donor characteristics (n=542)
| Histological Lesions | Donor Age | Stroke as Reason of Donor Death | Donor History of Hypertension | Donor Diabetes Mellitus | Donor History of Smoking |
|---|---|---|---|---|---|
| Interstitial fibrosis | |||||
| r | 0.29 | 0.11 | 0.17 | 0.01 | −0.01 |
| P | <0.001 | 0.01 | 0.0003 | 0.80 | 0.80 |
| Tubular atrophy | |||||
| r | 0.38 | 0.2 | 0.22 | −0.02 | 0.02 |
| P | <0.001 | <0.001 | 0.001 | 0.70 | 0.70 |
| Arteriolar hyalinosis | |||||
| r | 0.11 | 0.13 | 0.12 | 0.04 | 0.04 |
| P | 0.01 | 0.003 | 0.01 | 0.43 | 0.40 |
| Vascular intimal thickening | |||||
| r | 0.09 | 0.02 | 0.1 | 0.34 | 0.08 |
| P | 0.04 | 0.70 | 0.02 | 0.003 | 0.05 |
| Glomerulosclerosis | |||||
| r | 0.25 | 0.12 | 0.18 | 0.10 | 0.03 |
| P | <0.001 | 0.01 | 0.001 | 0.02 | 0.50 |
To further elucidate the complex relations between donor demographics and baseline histology, principal component analysis was applied. The summery plot is presented in Figure 3. This analysis illustrates the following dichotomy between the histologic lesions: (1) closely clustering lesions associating with donor age (glomerulosclerosis, tubular atrophy, interstitial fibrosis); and (2) other lesions (vascular intimal thickening and arteriolar hyalinosis) that correlated less with donor age, but better with history of smoking, diabetes mellitus, and hypertension.
Figure 3.
Principal component analysis using the individual histologic lesions in baseline biopsies and the different donor demographics of 548 renal allograft recipients. This analysis illustrates the following dichotomy between the histologic lesions: (1) closely clustering lesions associating with donor age, such as glomerulosclerosis (gs), tubular atrophy (ct), and interstitial fibrosis (ci); and (2) other lesions, such as vascular intimal thickening (cv) and arteriolar hyalinosis (ah), that correlate less with donor age, but better with history of smoking, diabetes mellitus (DM), and hypertension (AHT) (also see Table 4). PRIN1, principal component 1; PRIN2, principal component 2.
Histologic Scoring Systems for Baseline Biopsies and Prediction of Graft Survival
As was expected from the significant association between the individual histologic lesions and death-censored graft survival, there was a highly significant association between each of the previously published histologic scoring systems and death-censored graft survival in Cox proportional hazards analysis. The score proposed by Lopes et al.14 best associated with graft outcome (Figure 4 and Table 5). The area under the receiver operating characteristic (ROC) curve was largest for the score proposed by Remuzzi for prediction of 3-, 5-, and 10-year graft survival (Table 5). None of the previously proposed scoring systems, however, reached an area under the ROC curve higher than 0.62.
Figure 4.
Association of death-censored graft survival with the previously proposed Remuzzi score,7 the total chronic Banff score,16 and the donor chronic damage score,14, using Kaplan–Meier survival analysis, and the corresponding ROC curves for prediction of (death-censored) 5-year graft survival. TX, transplantation.
Table 5.
Predictive performance of the previously proposed histologic scoring systems for death-censored graft survival, in all donor kidneys (n=542)
| Scoring System | Cox Proportional Hazards Analysis | Logistic Regression | |||||||
|---|---|---|---|---|---|---|---|---|---|
| 3-yr Survival | 5-yr Survival | 10-yr Survival | |||||||
| Comparison | HR (95% CI) | P | AUC under ROC Curve | P | AUC under ROC curve | P | AUC under ROC curve | P | |
| Remuzzi score7 | ≥7 versus <7 | 1.22 (1.09 to 1.38) | 0.001 | 0.62 | 0.02 | 0.62 | 0.003 | 0.59 | 0.09 |
| Donor chronic damage score14 | ≥3 versus <3 | 1.52 (1.23 to 1.88) | 0.001 | 0.55 | 0.29 | 0.59 | 0.03 | 0.59 | 0.08 |
| Total chronic Banff score16 | ≥6 versus <6 | 1.25 (1.1 to 1.42) | 0.0004 | 0.6 | 0.05 | 0.6 | 0.01 | 0.59 | 0.08 |
Creation and Validation of a New Scoring System for Implantation Biopsies
Because the previously proposed histologic scoring systems did not provide sufficient predictive performance in our cohort, we created an alternative scoring system (Leuven donor risk score), based on both histologic data and donor age. The reason for including donor age in this clinicopathologic scoring system was the finding that donor age was associated with death-censored graft survival, independent of the baseline graft histology, and because including donor age in decisions about kidney graft acceptance or discard is clinically reasonable. This scoring system was created by logistic regression analyses, including the interstitial fibrosis/tubular atrophy (IFTA) grade and glomerulosclerosis score (GS) as histologic parameters and donor age as the clinical parameter. Donor age was entered in the analysis as a continuous variable. The clinicopathologic scoring system was created in the historic cohort (n=181 for prediction of 3- and 5-year survival; n=124 for prediction of 10-year survival) for prediction of 3-year, 5-year, and 10-year overall graft survival (Table 6). The predictive performance of this novel scoring system was then tested in the validation cohort (n=367 for prediction of 3-year survival; n=145 for prediction of 5-year survival; 10-year survival data are not available in the validation cohort).
Table 6.
Predictive performance of the newly created clinicopathologic Leuven donor risk score
| Score | Algorithm Obtained from Logistic Regression Analysis on Historic Cohort | Historic Cohort | Validation Cohort | ||
|---|---|---|---|---|---|
| AUC under ROC Curve | P | AUC under ROC Curve | P | ||
| 1 (3-yr prediction) | Age + 2 GS + 10 IFTA | 0.65 | 0.03 | 0.70 | 0.01 |
| 2 (Leuven donor risk score; 5-yr prediction) | GS + 3 IFTA + age | 0.67 | 0.002 | 0.81 | <0.001 |
| 3 (10-yr prediction) | 1.5 age + 14 GS + IFTA | 0.60 | 0.05 | – | – |
Logistic regression analysis was performed on the historic cohort (n=181 for 3- and 5-year survival; n=124 for 10-year survival) to construct a predictive algorithm for graft survival at 3, 5, and 10 years after transplantation. This algorithm was then applied to the validation cohort (n=367 for 3-year survival; n=145 for 5-year survival). gs, glomerulosclerosis score (gs <10%=0; gs >10%=1); ifta, interstitial fibrosis/tubular atrophy grade according to Banff classification.
As was predicted from the univariate association between the individual parameters and graft survival, all three newly created scores (for prediction of 3-, 5-, or 10-year survival) were highly significantly associated with graft survival. ROC analysis on the validation cohort showed area under the curve (AUC) values of 0.70 for score 1 (age + 2 GS + 10 IFTA [3-year survival]) (P=0.01) and 0.81 for score 2 (age + GS + 3 IFTA [5-year survival]) (P<0.001) (Figure 5 and Table 6). On the basis of the best performing algorithm (score 2: age + GS + 3 IFTA), a Leuven donor risk score >47 has 85% specificity and 81% sensitivity for graft failure within the first 5 years after transplantation. When 60 were used as a cut-off, specificity rose to 90%, with a sensitivity of 66% (Figure 5).
Figure 5.
Statistical evaluation of the Leuven donor risk score. (A) Association of death-censored graft survival with the newly constructed Leuven donor risk score, using Kaplan–Meier survival analysis. (B) ROC curves for prediction of (death-censored) 5-year graft survival based on the Leuven donor risk score in the historic cohort (n=181) and in the validation cohort (n=367). TX, transplantation.
Performance of the Leuven Donor Risk Score on Frozen Sections
Finally, we evaluated the applicability of the Leuven donor risk score on frozen sections, because allocation decisions can only be based on histology if fast sample processing yields sufficiently reliable results. The correlation between frozen sections and paraffin-embedded sections was excellent for the individual histologic lesions (arteriolar hyalinosis, intimal fibrosis, tubular atrophy, and glomerulosclerosis; all P<0.001, r=0.75–0.87), for ifta grade (P<0.001, r=0.95), and for the Leuven donor risk score (P<0.001, r=0.99).
Discussion
In this study, we evaluated the effect of baseline histologic lesions and donor characteristics on renal allograft survival, and demonstrated that global glomerulosclerosis and interstitial fibrosis/tubular atrophy were significantly associated with impaired graft outcome, independent of baseline donor demographics such as donor age and stroke as the cause of donor death. Other lesions, such as arteriolar hyalinosis and vascular intimal thickening (which were closely related to donor history of hypertension, stroke, diabetes mellitus, and smoking), did not associate with graft survival. None of the histologic scoring systems previously proposed by other groups7,14,16 had sufficient prognostic value when applied to our cohort, whereas a newly created clinicopathologic algorithm for prediction of graft outcome (Leuven donor risk score) provided better predictive accuracy to be used for donor kidney allocation.
Previous studies revealed that donor age and the histology of the graft at the time of transplantation are major determinants of graft function. Anglicheau et al. showed that glomerulosclerosis at the time of transplantation predicts low estimated GFR at 1 year after transplant.8 Arias et al. also demonstrated that glomerulosclerosis, interstitial fibrosis, tubular atrophy, and arteriolar hyalinosis in implantation biopsies predicted estimated GFR at 3 years after transplant.9 Although previous studies showed that post-transplant renal function in the first year predicts long-term kidney transplant survival,18 graft function at 1 year is unlikely to be the major endpoint that needs to be correlated with baseline findings to decide whether to accept or discard a donor graft. Therefore, studies correlating baseline histology with graft survival serve as a better basis for guiding kidney allocation.
Importantly, in this study, baseline biopsies were not used to refuse a donor kidney. This study thus represents an ideal setting to study the predictive performance of baseline biopsies for graft outcome. The validity of previous studies is indeed hampered by the inherent selection bias present in these studies, because the clinical transplant centers that analyzed the predictive capacity of baseline histology for graft survival discard kidneys based on histologic lesions,5 and can thus only evaluate histologic risk factors in kidneys that were actually selected for transplantation.
In this study, the predictive performance of previously published and widely used histologic scoring systems, as well as the newly proposed Leuven donor risk score, was insufficient to use any of these scoring systems for kidney discard or acceptance. Notwithstanding this, we demonstrated that it is possible to create a scoring system with sufficient predictive performance to guide clinical decisions at the time of transplantation (kidney graft allocation). Furthermore, we proved the validity of the Leuven donor risk score in frozen sections as well. This is important because fast procedures are essential to incorporate histologic scoring in clinical decision algorithms with strict timing constraints, such as donor kidney allocation. It remains to be evaluated whether the reproducibility of the proposed Leuven donor risk score remains as high in routine clinical practice as it was in this study.
Previously, application of the Remuzzi score was proposed for deciding between single or dual transplantation, and proved to be beneficial for graft outcome.6 This study clearly demonstrated the benefit of dual kidney transplantation in the case of a high degree of chronic damage in the pretransplant biopsy. In our cohort, dual transplantation was not routinely performed. Moreover, only a minority of our donors were aged >60 years, and only a small proportion of our baseline biopsies had a Remuzzi score >7. Therefore, the approach proposed by Remuzzi et al., to use baseline biopsy histology for deciding between dual and single kidney transplantation and in the case of a score >7 to discard a kidney, cannot be evaluated in this study. However, our study shows that implantation biopsy evaluation is not only relevant for older donor kidneys. Scarred kidneys from young donors also represent a significantly increased risk for premature graft loss. To improve the overall outcome of the transplant population, it could be beneficial to include baseline histology in the allocation algorithm for these younger donors. Because implementation of additional parameters in any allocation algorithm will have a certain effect on the kidney transplant wait list and on the equity of access to transplantation, it is beyond the scope of this study to calculate the potential effect of implementation of the Leuven donor risk score in a kidney allocation algorithm. First, validation of our findings in an independent and larger cohort is necessary. Second, proper computer simulations in very large data sets should calculate the effect on patient and graft outcome of any change in the allocation system.
In this study, we demonstrated that arteriolar hyalinosis and vascular intimal thickening in baseline biopsies did not associate with impaired long-term allograft survival. This finding contrasts with previous studies that showed that vascular luminal narrowing >25%13 and arteriolar hyalinosis15 are independent predictors of graft failure. Four other studies, however, did not find any association of vascular intimal thickening and arteriolar hyalinosis with graft function at 12 months.11–13 The reason for these discrepancies could be found in the relatively low number of baseline biopsies with vascular intimal thickening in this study, and selection bias by kidney discard based on these lesions in other studies (which was not present in this study). In addition, the study by Munivenkatappa et al. and the derivation of the MAPI15 were partly based on biopsies imported from other organ procurement organizations, after rejection of the kidneys by other local centers. This background could have also affected the results of this previous study.
Clearly, our findings are subject to several limitations. Not all patients were treated with the same immunosuppressive regimen and patients included in the historic cohort received an outdated regimen. However, the results were comparable when the scoring systems were applied in a more recent validation cohort that was treated with modern immunosuppression and was thus more representative of current clinical practice. In addition, although donor clinical demographics were collected prospectively, many parameters that could be of importance are often recorded incompletely. In addition, the specific setting of deceased donor donation also leads to underreporting of medical risk factors such as smoking status and hypertension. Vascular intimal thickening was relatively rare in this study, and low numbers could partially be responsible for absence of association of this parameter with graft outcome. Furthermore, an important drawback to compare our data with previous data is that we were not able to evaluate the MAPI score15 in our biopsy cohort. The ROC curve performed on the Maryland validation cohort had an AUC of 0.74, which is less than the performance of the newly created scores tested in our validation cohort. However, application of the MAPI score at the time of transplantation is cumbersome in many transplant centers, because morphometric analysis is not universally available (especially not in an urgent clinical setting). Finally, it has to be acknowledged that the reproducibility of histologic scoring is known to be moderate at best,19 and that sampling error is inherent to any biopsy study. For this reason, all biopsies were obtained in a comparable fashion (wedge biopsies), and one single pathologist, blinded for any clinical information, rescored all biopsies retrospectively. Despite these measures, sampling error and intraobserver variability are still possible, and could account for the absence of association between arteriolar hyalinosis and vascular intimal thickening, which often have a patchy distribution and are known the be less reproducible than the other Banff qualifiers.20 In addition, the use of wedge biopsies in our cohort could be a possible explanation for the lack of significant association between vascular intimal thickening and graft outcome in this study, because it was shown previously that core biopsies are superior to wedge biopsies to evaluate this lesion.21 A standardized, more objective evaluation of biopsies, either by morphometric analysis, or by molecular profiling, could yield a better tool for assessment of organ quality and prediction of organ survival after transplantation.22–24
In conclusion, our study demonstrates that donor age and the histology of the graft at the time of transplantation are major determinants of long-term graft outcome. In our center, baseline biopsies are not systematically used for donor kidney allocation. The current cohorts thus represent an ideal study cohort, because selection bias is avoided. Based on the finding that age-associated histologic lesions (interstitial fibrosis/tubular atrophy and glomerulosclerosis) are highly significant predictors of survival, we have created a novel scoring system with these parameters. The predictive performance of the previously proposed scoring systems proved to be poor in our cohort. In contrast, our newly constructed scoring system provided sufficient predictive performance to guide kidney allocation.
Concise Methods
Patient Inclusion
Between January 1991 and June 2009, 2163 kidney transplants were performed in adult patients at the University Hospitals Leuven. Protocol baseline biopsies have been increasingly introduced at our center not to guide acceptance or discard but to provide a reference point in case of post-transplant dysfunction. Between 1991 and 2003 (transition period), baseline biopsies were performed not systematically but only at the request of the transplant surgeon in the case of donor demographic variables that were perceived to be associated with post-transplant outcome. These biopsies were not systematically used for allocation purposes or for kidney discard, but as benchmark for comparison of post-transplantation histology in case a post-transplant indication biopsy was necessary because of graft dysfunction. Since 2004, baseline biopsies were systematically performed for post-transplant comparison, again without pretransplant histologic assessment. All patients who underwent a baseline biopsy at the time of transplantation were included in this study. To build and validate a histologic scoring system in baseline biopsies, the cohort was divided in a historic cohort (transplantation before 2004) and a validation cohort (transplantation after 2004).
Biopsies and Histologic Evaluation
Biopsies were performed after arrival of the kidney at our center and during the bench work and preparation of the kidney. One experienced renal pathologist (E.L.) retrospectively reviewed all baseline biopsies, blinded for clinical data. The biopsy specimens were wedge biopsies, with slides containing 4–10 paraffin sections (2 μm) that were stained with hematoxylin and eosin, periodic acid–Schiff, and a silver methenamine staining method (Jones). The severity of chronic histologic lesions (interstitial fibrosis, tubular atrophy, arteriolar hyalinosis, vascular intimal thickening) was scored semiquantitatively according to the revised Banff criteria, as well as the ifta grade.25 In addition, the total number and the number of sclerosed glomeruli were counted. Finally, frozen sections of 30 biopsies were stained with hematoxylin and eosin, and independently rescored for correlation analysis with the results obtained from the paraffin-embedded sections.
Clinical Data Collection
Patient and donor data were prospectively collected in electronic clinical patient charts. The clinical database was transferred to SAS data files (SAS Institute, Cary, NC) at the time of analysis. The following data were collected: donor characteristics (age, sex, cause of death, history of hypertension of diabetes, serum creatinine levels before organ recovery, extended criteria donation), cold ischemia time, and recipient characteristics at the time of transplantation (age, sex, number of previous kidney transplantations, preexisting kidney disease, number of HLA mismatches). Graft outcome data were also recorded. Delayed graft function was defined as the need for dialysis within the first week after transplantation. Graft failure was defined as either return to hemodialysis, transplant nephrectomy, or recipient death with a functioning graft. Death-censored graft survival was calculated as the time from transplantation to graft nephrectomy or return to dialysis, with time from transplantation to death of the recipient with a functioning graft as the censoring variable. The definition of extended criteria donation included all donors aged ≥60 years and donors aged 50–59 years with at least two of three other conditions (cerebrovascular cause of death, terminal creatinine >1.5 mg/dl, and hypertension).
Statistical Analyses
Histologic covariates entered in the univariate analysis included interstitial fibrosis (Banff ci score), tubular atrophy (Banff ct score), arteriolar hyalinosis (Banff ah score), vascular intimal thickening (Banff cv score), and glomerulosclerosis. Donor demographics entered in the univariate analysis included history of hypertension, diabetes, smoking, age, and cerebrovascular cause of death. All covariates associated at the P=0.20 level with death-censored graft survival in univariate analysis were included in the multivariate analysis. The multivariate Cox proportional hazards analysis included only recipients with complete data on graft survival, donors’ and recipients’ age and sex, donor source (living or deceased) and cause of death, number of HLA mismatches, original kidney disease, primary immunosuppressive regimen, and year of transplantation. In addition, the explanatory effect of the donor demographics on the histologic lesions was analyzed. Final Cox proportional hazards regression models were constructed by backward variable selection. For illustrations, Kaplan–Meier survival curves with log-rank testing were used.
From the individual Banff scores that were recorded for every biopsy, previously published histologic scores were calculated. The Remuzzi score was calculated by the addition of four different parameters: glomerular global sclerosis (0–3), tubular atrophy (0–3), interstitial fibrosis (0–3), and arterial and arteriolar narrowing (0–3).6 The donor chronic damage score was calculated by the addition of the glomerulosclerosis score (0 = glomerulosclerosis <10%; 1 = glomerulosclerosis >10%), the vascular intimal thickening score, and the interstitial fibrosis score.14 The total chronic Banff score was calculated as the addition of the scores for transplant glomerulopathy, mesangial matrix increase, tubular atrophy, interstitial fibrosis, vascular intimal thickening, arteriolar hyalinosis, and the [fraction sclerosed glomeruli] × 3.16 The MAPI15 was not calculated because of the need for morphometric biopsy analysis.
In the historic cohort, a new pathologic scoring system was created with the histologic lesions and donor demographics that were significant in multivariate Cox proportional hazards analyses for association with graft survival. The new scoring system was constructed using logistic regression analysis, with death-censored graft survival at 3, 5, and 10 years after transplantation as dependent variables. ROC analysis was used to evaluate the predictive performance of the newly created scoring system in the validation cohort.
For variance analysis of continuous variables in different groups, the nonparametric Wilcoxon-Mann-Whitney U test, nonparametric ANOVA, and parametric one-way ANOVA were used, as appropriate. Dichotomous variables were compared using the chi-squared test. For evaluation of the correlation of the different histologic lesions, Spearman correlation analysis was used, as well as principal component analysis. From the principal component analysis, a two-dimensional scatter plot was generated to represent the histologic lesions according to their score in the loading matrix. To evaluate the validity of the histologic scoring in frozen sections, the separate histologic lesions, ifta grade, and the newly created score calculated from data obtained on frozen sections were correlated with the data obtained from paraffin-embedded sections by using Spearman correlation analysis.
The results are expressed as numerical values and percentages for categorical variables and as means ± SDs for continuous variables, unless stated otherwise. All tests were two-sided and P values of <0.05 were considered to be statistically significant. Analyses were performed with SAS (version 9.2; SAS Institute), JMP9.0 (SAS Institute), and GraphPad Prism (version 5.00; GraphPad Software, San Diego, CA) software.
Disclosures
None.
Acknowledgments
We thank the centers of the Leuven Collaborative Group for Renal Transplantation, as well as the clinicians, surgeons, nursing staff, and patients.
The authors had full access to the data and take responsibility for its integrity. All authors have read and agree with the manuscript as written.
Footnotes
Published online ahead of print. Publication date available at www.jasn.org.
See related editorial, “Foretelling the Future: Predicting Graft Outcome by Evaluating Kidney Baseline Transplant Biopsies,” on pages 1716–1719.
References
- 1.Port FK, Merion RM, Roys EC, Wolfe RA: Trends in organ donation and transplantation in the United States, 1997-2006. Am J Transplant 8: 911–921, 2008 [DOI] [PubMed] [Google Scholar]
- 2.Meier-Kriesche HU, Schold JD, Srinivas TR, Kaplan B: Lack of improvement in renal allograft survival despite a marked decrease in acute rejection rates over the most recent era. Am J Transplant 4: 378–383, 2004 [DOI] [PubMed] [Google Scholar]
- 3.Summers DM, Johnson RJ, Allen J, Fuggle SV, Collett D, Watson CJ, Bradley JA: Analysis of factors that affect outcome after transplantation of kidneys donated after cardiac death in the UK: A cohort study. Lancet 376: 1303–1311, 2010 [DOI] [PubMed] [Google Scholar]
- 4.Opelz G, Döhler B, Collaborative Transplant Study : Influence of immunosuppressive regimens on graft survival and secondary outcomes after kidney transplantation. Transplantation 87: 795–802, 2009 [DOI] [PubMed] [Google Scholar]
- 5.Jochmans I, Pirenne J: Graft quality assessment in kidney transplantation: Not an exact science yet! Curr Opin Organ Transplant 16: 174–179, 2011 [DOI] [PubMed] [Google Scholar]
- 6.Remuzzi G, Grinyò J, Ruggenenti P, Beatini M, Cole EH, Milford EL, Brenner BM, Double Kidney Transplant Group (DKG) : Early experience with dual kidney transplantation in adults using expanded donor criteria. J Am Soc Nephrol 10: 2591–2598, 1999 [DOI] [PubMed] [Google Scholar]
- 7.Remuzzi G, Cravedi P, Perna A, Dimitrov BD, Turturro M, Locatelli G, Rigotti P, Baldan N, Beatini M, Valente U, Scalamogna M, Ruggenenti P, Dual Kidney Transplant Group : Long-term outcome of renal transplantation from older donors. N Engl J Med 354: 343–352, 2006 [DOI] [PubMed] [Google Scholar]
- 8.Anglicheau D, Loupy A, Lefaucheur C, Pessione F, Létourneau I, Côté I, Gaha K, Noël LH, Patey N, Droz D, Martinez F, Zuber J, Glotz D, Thervet E, Legendre C: A simple clinico-histopathological composite scoring system is highly predictive of graft outcomes in marginal donors. Am J Transplant 8: 2325–2334, 2008 [DOI] [PubMed] [Google Scholar]
- 9.Arias LF, Blanco J, Sanchez-Fructuoso A, Prats D, Duque E, Sáiz-Pardo M, Ruiz J, Barrientos A: Histologic assessment of donor kidneys and graft outcome: Multivariate analyses. Transplant Proc 39: 1368–1370, 2007 [DOI] [PubMed] [Google Scholar]
- 10.Escofet X, Osman H, Griffiths DF, Woydag S, Adam Jurewicz W: The presence of glomerular sclerosis at time zero has a significant impact on function after cadaveric renal transplantation. Transplantation 75: 344–346, 2003 [DOI] [PubMed] [Google Scholar]
- 11.Randhawa PS, Minervini MI, Lombardero M, Duquesnoy R, Fung J, Shapiro R, Jordan M, Vivas C, Scantlebury V, Demetris A: Biopsy of marginal donor kidneys: Correlation of histologic findings with graft dysfunction. Transplantation 69: 1352–1357, 2000 [DOI] [PubMed] [Google Scholar]
- 12.Pokorná E, Vítko S, Chadimová M, Schück O, Ekberg H: Proportion of glomerulosclerosis in procurement wedge renal biopsy cannot alone discriminate for acceptance of marginal donors. Transplantation 69: 36–43, 2009 [DOI] [PubMed] [Google Scholar]
- 13.Kayler LK, Mohanka R, Basu A, Shapiro R, Randhawa PS: Correlation of histologic findings on preimplant biopsy with kidney graft survival. Transpl Int 21: 892–898, 2008 [DOI] [PubMed] [Google Scholar]
- 14.Lopes JA, Moreso F, Riera L, Carrera M, Ibernon M, Fulladosa X, Grinyó JM, Serón D: Evaluation of pre-implantation kidney biopsies: Comparison of Banff criteria to a morphometric approach. Kidney Int 67: 1595–1600, 2005 [DOI] [PubMed] [Google Scholar]
- 15.Munivenkatappa RB, Schweitzer EJ, Papadimitriou JC, Drachenberg CB, Thom KA, Perencevich EN, Haririan A, Rasetto F, Cooper M, Campos L, Barth RN, Bartlett ST, Philosophe B: The Maryland aggregate pathology index: A deceased donor kidney biopsy scoring system for predicting graft failure. Am J Transplant 8: 2316–2324, 2008 [DOI] [PubMed] [Google Scholar]
- 16.Snoeijs MG, Buurman WA, Christiaans MH, van Hooff JP, Goldschmeding R, van Suylen RJ, Peutz-Kootstra CJ, van Heurn LW: Histological assessment of preimplantation biopsies may improve selection of kidneys from old donors after cardiac death. Am J Transplant 8: 1844–1851, 2008 [DOI] [PubMed] [Google Scholar]
- 17.Woestenburg AT, Verpooten GA, Ysebaert DK, Van Marck EA, Verbeelen D, Bosmans JL: Fibrous intimal thickening at implantation adversely affects long-term kidney allograft function. Transplantation 87: 72–78, 2009 [DOI] [PubMed] [Google Scholar]
- 18.Hariharan S, McBride MA, Cherikh WS, Tolleris CB, Bresnahan BA, Johnson CP: Post-transplant renal function in the first year predicts long-term kidney transplant survival. Kidney Int 62: 311–318, 2002 [DOI] [PubMed] [Google Scholar]
- 19.Furness PN, Taub N, Assmann KJ, Banfi G, Cosyns JP, Dorman AM, Hill CM, Kapper SK, Waldherr R, Laurinavicius A, Marcussen N, Martins AP, Nogueira M, Regele H, Seron D, Carrera M, Sund S, Taskinen EI, Paavonen T, Tihomirova T, Rosenthal R: International variation in histologic grading is large, and persistent feedback does not improve reproducibility. Am J Surg Pathol 27: 805–810, 2003 [DOI] [PubMed] [Google Scholar]
- 20.Sis B, Dadras F, Khoshjou F, Cockfield S, Mihatsch MJ, Solez K: Reproducibility studies on arteriolar hyaline thickening scoring in calcineurin inhibitor-treated renal allograft recipients. Am J Transplant 6: 1444–1450, 2006 [DOI] [PubMed] [Google Scholar]
- 21.Haas M, Segev DL, Racusen LC, Bagnasco SM, Melancon JK, Tan M, Kraus ES, Rabb H, Ugarte RM, Burdick JF, Montgomery RA: Arteriosclerosis in kidneys from healthy live donors: comparison of wedge and needle core perioperative biopsies. Arch Pathol Lab Med 132: 37–42, 2008 [DOI] [PubMed] [Google Scholar]
- 22.Kainz A, Mitterbauer C, Hauser P, Schwarz C, Regele HM, Berlakovich G, Mayer G, Perco P, Mayer B, Meyer TW, Oberbauer R: Alterations in gene expression in cadaveric vs. live donor kidneys suggest impaired tubular counterbalance of oxidative stress at implantation. Am J Transplant 4: 1595–1604, 2004 [DOI] [PubMed] [Google Scholar]
- 23.Mueller TF, Reeve J, Jhangri GS, Mengel M, Jacaj Z, Cairo L, Obeidat M, Todd G, Moore R, Famulski KS, Cruz J, Wishart D, Meng C, Sis B, Solez K, Kaplan B, Halloran PF: The transcriptome of the implant biopsy identifies donor kidneys at increased risk of delayed graft function. Am J Transplant 8: 78–85, 2008 [DOI] [PubMed] [Google Scholar]
- 24.Naesens M, Li L, Ying L, Sansanwal P, Sigdel TK, Hsieh SC, Kambham N, Lerut E, Salvatierra O, Butte AJ, Sarwal MM: Expression of complement components differs between kidney allografts from living and deceased donors. J Am Soc Nephrol 20: 1839–1851, 2009 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Solez K, Colvin RB, Racusen LC, Haas M, Sis B, Mengel M, Halloran PF, Baldwin W, Banfi G, Collins AB, Cosio F, David DS, Drachenberg C, Einecke G, Fogo AB, Gibson IW, Glotz D, Iskandar SS, Kraus E, Lerut E, Mannon RB, Mihatsch M, Nankivell BJ, Nickeleit V, Papadimitriou JC, Randhawa P, Regele H, Renaudin K, Roberts I, Seron D, Smith RN, Valente M: Banff 07 classification of renal allograft pathology: Updates and future directions. Am J Transplant 8: 753–760, 2008 [DOI] [PubMed] [Google Scholar]