Living donors with low iGFR: does this put two donors at risk?
John Asher, Colin Geddes, Siobhan McManus, Marc Clancy
02/03/2020
Supplementary data for BTS 2020 poster P37
Total of 380 living kidney donors worked up in Glasgow since January 2009 included in this analysis. 12 had isotopic GFR of 60-69 ml/min/1.73m², 170 between 70 and 89, and 198 over 90.
Basic donor characteristics
Donor age
The observed differences between the three groups for donor ages were statistically significant. This is not surprising given the different acceptable GFR thresholds for age in the UK guidelines.
Df Sum Sq Mean Sq F value Pr(>F)
GFR_group 2 10831 5416 53.35 <2e-16 ***
Residuals 377 38267 102
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1Donor blood pressure
Systolic BP
In univariate and multivariate comparisons, there was no association between iGFR and pre-donation systolic BP, although donor age was a significant predictor of systolic BP.
Df Sum Sq Mean Sq F value Pr(>F)
GFR_group 2 1124 561.8 2.273 0.105
Residuals 342 84513 247.1
35 observations deleted due to missingnessMultivariate analysis:
Anova Table (Type III tests)
Response: Preop_sys_BP
Sum Sq Df F value Pr(>F)
(Intercept) 52047 1 229.9665 < 2.2e-16 ***
Age 7337 1 32.4188 2.683e-08 ***
GFR_group 123 2 0.2706 0.7631
Residuals 77176 341
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1Effect sizes by \(\eta^2\) in the multivariate analysis:
term etasq
1 Age 0.097
2 GFR_group 0.001Diastolic BP
In univariate ANOVA, there was an apparent association between donor iGFR and diastolic BP, but it was not statistically significant and no pairwise comparisons were even close to significant on post-hoc testing.
Df Sum Sq Mean Sq F value Pr(>F)
GFR_group 2 404 201.85 2.585 0.0768 .
Residuals 342 26703 78.08
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
35 observations deleted due to missingness Tukey multiple comparisons of means
95% family-wise confidence level
Fit: aov(formula = Preop_dia_BP ~ GFR_group, data = df)
$GFR_group
diff lwr upr p adj
70-89-60-69 -2.850000 -9.635071 3.9350714 0.5844666
90+-60-69 -4.627933 -11.386694 2.1308277 0.2419889
90+-70-89 -1.777933 -4.056156 0.5002902 0.1591310In the multivariate ANCOVA, there was no association between iGFR and pre-donation diastolic BP, although donor age was a significant predictor of diastolic BP.
Anova Table (Type III tests)
Response: Preop_dia_BP
Sum Sq Df F value Pr(>F)
(Intercept) 21622.5 1 291.4054 < 2.2e-16 ***
Age 1400.0 1 18.8677 1.85e-05 ***
GFR_group 1.4 2 0.0092 0.9908
Residuals 25302.5 341
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1Effect sizes:
term etasq
1 Age 0.066
2 GFR_group 0.000Urinary protein
Proteinuria by lab measurement
There was no statistically significant differences between iGFR groups in last predonation urinary protein.
Df Sum Sq Mean Sq F value Pr(>F)
GFR_group 2 0.0316 0.015811 2.323 0.101
Residuals 173 1.1776 0.006807
204 observations deleted due to missingnessThere were also no statistically significant differences in a multivariate analysis:
Anova Table (Type III tests)
Response: Predonation_UProtein
Sum Sq Df F value Pr(>F)
(Intercept) 0.04977 1 7.6974 0.006219 **
Preop_sys_BP 0.00053 1 0.0815 0.775599
Preop_dia_BP 0.00082 1 0.1263 0.722760
Age 0.00835 1 1.2920 0.257448
GFR_group 0.02877 2 2.2246 0.111591
Residuals 0.98933 153
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1Effect sizes:
term etasq
1 Preop_sys_BP 0.002
2 Preop_dia_BP 0.000
3 Age 0.006
4 GFR_group 0.028Albumin/creatinine ratio
There were no statistically significant differences in last predonation ACR.
Df Sum Sq Mean Sq F value Pr(>F)
GFR_group 2 5.4 2.707 0.461 0.631
Residuals 377 2214.1 5.873 There were also no statistically significant differences in a multivariate analysis:
Anova Table (Type III tests)
Response: Latest_preop_ACR
Sum Sq Df F value Pr(>F)
(Intercept) 0.41 1 0.0632 0.8016
Preop_sys_BP 0.67 1 0.1034 0.7480
Preop_dia_BP 0.04 1 0.0066 0.9355
Age 1.96 1 0.3037 0.5820
GFR_group 4.00 2 0.3099 0.7337
Residuals 2188.07 339 Effect sizes:
term etasq
1 Preop_sys_BP 0.000
2 Preop_dia_BP 0.000
3 Age 0.002
4 GFR_group 0.002Donor outcome
eGFR at 1 and 5 years post donation
ANCOVA for donor eGFR at 1 year
Multivariate analysis showed donor age and last preoperative diastolic BP were also statistically significant predictors of donor eGFR at 1 and 5 years from donation.
Anova Table (Type III tests)
Response: eGFR_1yr
Sum Sq Df F value Pr(>F)
(Intercept) 11992.2 1 92.0664 < 2.2e-16 ***
Age 1930.9 1 14.8238 0.0001516 ***
Preop_dia_BP 510.2 1 3.9166 0.0489583 *
GFR_group 2217.9 2 8.5138 0.0002681 ***
Residuals 31131.2 239
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1Effect sizes:
term etasq
1 Age 0.161
2 Preop_dia_BP 0.015
3 GFR_group 0.055Crude and adjusted mean eGFR at 1 year
The crude mean eGFR at 1 year is:
| GFR_group | Mean | SD | min | max |
|---|---|---|---|---|
| 60-69 | 50.6 | 4.8 | 44.3 | 59.9 |
| 70-89 | 62.3 | 11.1 | 35.5 | 84.9 |
| 90+ | 70.8 | 12.6 | 45.0 | 108.3 |
The mean donor eGFR at 1 year adjusted for the covariates of donor age and pre-donation diastolic BP (“fit”), together with standard error (“se”) and 95% confidence intervals (“lower” and “upper”), is:
| GFR_group | fit | se | lower | upper |
|---|---|---|---|---|
| 60-69 | 55.4 | 4.2 | 47.2 | 63.6 |
| 70-89 | 63.5 | 1.1 | 61.3 | 65.7 |
| 90+ | 69.2 | 1.1 | 67.0 | 71.4 |
ANCOVA for donor eGFR at 5 years
Interestingly, the donor iGFR group was not a signficant predictor of donor eGFR at 5 years when adjusted for covariates.
Anova Table (Type III tests)
Response: eGFR_5yr
Sum Sq Df F value Pr(>F)
(Intercept) 10847.2 1 60.1657 2.988e-11 ***
Age 3597.9 1 19.9561 2.680e-05 ***
Preop_dia_BP 573.8 1 3.1825 0.07837 .
GFR_group 474.5 2 1.3160 0.27418
Residuals 13882.3 77
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1Effect sizes:
term etasq
1 Age 0.349
2 Preop_dia_BP 0.025
3 GFR_group 0.021Crude and adjusted mean eGFR at 5 years
The crude mean eGFR at 1 year is:
| GFR_group | Mean | SD | min | max |
|---|---|---|---|---|
| 60-69 | 53.5 | 8.1 | 43.6 | 65.6 |
| 70-89 | 62.4 | 15.9 | 13.1 | 115.4 |
| 90+ | 74.0 | 13.8 | 50.1 | 114.3 |
The mean donor eGFR at 5 years adjusted for the covariates of donor age and pre-donation diastolic BP (“fit”), together with standard error (“se”) and 95% confidence intervals (“lower” and “upper”), is:
| GFR_group | fit | se | lower | upper |
|---|---|---|---|---|
| 60-69 | 65.8 | 6.5 | 52.9 | 78.6 |
| 70-89 | 66.2 | 2.4 | 61.4 | 71.0 |
| 90+ | 71.6 | 2.2 | 67.3 | 75.9 |
Effect of donor iGFR on post-donation BP
Post-donation systolic BP
There were no significant differences in most recent post-donation systolic BP:
Df Sum Sq Mean Sq F value Pr(>F)
GFR_group 2 12661 6330 1.812 0.165
Residuals 370 1292850 3494
7 observations deleted due to missingnessMultivariate analysis showed that the GFR group was not a significant predictor of postdonation systolic BP, but pre-donation systolic BP was almost a significant predictor:
Anova Table (Type III tests)
Response: Latest_sys_BP
Sum Sq Df F value Pr(>F)
(Intercept) 7409 1 1.9947 0.15876
Age 1017 1 0.2738 0.60116
Preop_sys_BP 13752 1 3.7027 0.05516 .
Preop_dia_BP 1440 1 0.3877 0.53393
GFR_group 9993 2 1.3453 0.26185
Residuals 1259076 339
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1Although predonation systolic BP was almost a significant predictor of postdonation systolic BP, it can be seen to have only a small influence (\(\eta^2\)=0.018) in the effect sizes:
term etasq
1 Age 0.003
2 Preop_sys_BP 0.018
3 Preop_dia_BP 0.001
4 GFR_group 0.008Post-donation diastolic BP
There were significant differences in most recent post-donation diastolic BP:
Df Sum Sq Mean Sq F value Pr(>F)
GFR_group 2 656 327.8 3.395 0.0346 *
Residuals 370 35726 96.6
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
7 observations deleted due to missingnessOnly the 70-89 vs 90+ pairwise comparison was significant on post-hoc testing:
Tukey multiple comparisons of means
95% family-wise confidence level
Fit: aov(formula = Latest_dia_BP ~ GFR_group, data = df)
$GFR_group
diff lwr upr p adj
70-89-60-69 2.1136364 -4.799904 9.0271768 0.7521481
90+-60-69 -0.5816327 -7.458007 6.2947412 0.9783974
90+-70-89 -2.6952690 -5.138306 -0.2522319 0.0264545On multivariate analysis, only predonation BP was a significant predictor; both systolic and diastolic were significant predictors and had moderate effect size (\(\eta^2\) 0.139 and 0.095 respectively).
Anova Table (Type III tests)
Response: Latest_dia_BP
Sum Sq Df F value Pr(>F)
(Intercept) 2065.1 1 30.8587 5.618e-08 ***
Age 37.0 1 0.5532 0.4575313
Preop_sys_BP 890.8 1 13.3112 0.0003053 ***
Preop_dia_BP 2991.9 1 44.7084 9.436e-11 ***
GFR_group 90.7 2 0.6773 0.5086502
Residuals 22686.4 339
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1Effect sizes:
term etasq
1 Age 0.041
2 Preop_sys_BP 0.139
3 Preop_dia_BP 0.095
4 GFR_group 0.003Effect of donor iGFR on post-donation proteinuria
Postdonation laboratory proteinuria models
On univariate analysis, GFR group did predict postdonation proteinuria.
Df Sum Sq Mean Sq F value Pr(>F)
GFR_group 2 0.156 0.07817 3.145 0.0446 *
Residuals 284 7.059 0.02486
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
93 observations deleted due to missingnessOn mutivariate analysis, only predonation ACR and predonation proteinuria predicted postdonation ACR, with reasonable effect sizes (\(\eta^2\) 0.131 and 0.200 respectively).
Anova Table (Type III tests)
Response: Postdonation_UProtein
Sum Sq Df F value Pr(>F)
(Intercept) 0.03088 1 3.5771 0.06048 .
Latest_preop_ACR 0.15159 1 17.5629 4.698e-05 ***
Predonation_UProtein 0.35481 1 41.1062 1.722e-09 ***
Preop_sys_BP 0.01056 1 1.2234 0.27044
Preop_dia_BP 0.00211 1 0.2444 0.62178
GFR_group 0.01952 2 1.1307 0.32551
Residuals 1.31200 152
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1Effect sizes:
term etasq
1 Latest_preop_ACR 0.131
2 Predonation_UProtein 0.200
3 Preop_sys_BP 0.004
4 Preop_dia_BP 0.001
5 GFR_group 0.010Postdonation ACR models
On univariate analysis, GFR group did not predict postdonation ACR.
Df Sum Sq Mean Sq F value Pr(>F)
GFR_group 2 9 4.609 0.493 0.611
Residuals 377 3524 9.346 On mutivariate analysis, only predonation ACR predicted postdonation ACR, with a large effect (\(\eta^2\)=0.651).
Anova Table (Type III tests)
Response: Latest_ACR
Sum Sq Df F value Pr(>F)
(Intercept) 4.63 1 1.3415 0.2476
Latest_preop_ACR 2223.61 1 643.8366 <2e-16 ***
Preop_sys_BP 6.42 1 1.8587 0.1737
Preop_dia_BP 3.72 1 1.0777 0.2999
GFR_group 13.47 2 1.9497 0.1439
Residuals 1170.80 339
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1Effect sizes:
term etasq
1 Latest_preop_ACR 0.651
2 Preop_sys_BP 0.001
3 Preop_dia_BP 0.001
4 GFR_group 0.004Referral back to nephrology
The cumulative hazard plot for time to referral of donor back from GP to renal services for cause (reduced renal function, hypertension, proteinuria) shows that very few donors are referred back, including almost none from the low GFR group.
Effect of iGFR on donor survival
The Kaplan-Meier curve of donor survival, with confidence intervals shaded, shows that the donors with lower GFR had poorer survival but the difference did not meet statistical significance.
Age is of course a significant confounder, so multivariate Cox regression model performed including donor age and predonation diastolic BP. In this model, only age was a significant predictor, so a better performing model was repeated using only age and not BP.
Initial model:
Call:
coxph(formula = donor.surv ~ Age + Preop_dia_BP + GFR_group,
data = df)
coef exp(coef) se(coef) z p
Age 1.423e-01 1.153e+00 6.271e-02 2.270 0.0232
Preop_dia_BP -1.218e-02 9.879e-01 5.494e-02 -0.222 0.8246
GFR_group70-89 -1.984e+01 2.419e-09 1.295e+04 -0.002 0.9988
GFR_group90+ 1.095e+00 2.989e+00 1.231e+00 0.890 0.3735
Likelihood ratio test=13.24 on 4 df, p=0.01018
n= 345, number of events= 5
(35 observations deleted due to missingness)Improved model:
Call:
coxph(formula = donor.surv ~ Age + GFR_group, data = df)
coef exp(coef) se(coef) z p
Age 1.367e-01 1.146e+00 6.074e-02 2.251 0.0244
GFR_group70-89 -1.979e+01 2.536e-09 1.341e+04 -0.001 0.9988
GFR_group90+ 1.108e+00 3.029e+00 1.249e+00 0.887 0.3748
Likelihood ratio test=12.92 on 3 df, p=0.004807
n= 380, number of events= 5 Recipient outcomes
Recipient eGFR at 1 and 5 years post donation
Recipient eGFR models
eGFR at 1 year
Df Sum Sq Mean Sq F value Pr(>F)
GFR_group 2 10712 5356 5.71 0.00373 **
Residuals 267 250469 938
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
110 observations deleted due to missingnessA multivariate model including donor age and donor-recipient age mismatch showed that the recipient eGFR at 1 year was not predicted by the donor GFR:
Anova Table (Type III tests)
Response: Rec_eGFR_1yr
Sum Sq Df F value Pr(>F)
(Intercept) 67218 1 102.0428 < 2.2e-16 ***
Age 49768 1 75.5520 3.777e-16 ***
Age_mismatch 54898 1 83.3395 < 2.2e-16 ***
GFR_group 340 2 0.2583 0.7725
Residuals 174563 265
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1Effect sizes:
term etasq
1 Age 0.119
2 Age_mismatch 0.212
3 GFR_group 0.001| GFR_group | eGFR_1yr.mean | eGFR_1yr.sd | eGFR_5yr.mean | eGFR_5yr.sd |
|---|---|---|---|---|
| 60-69 | 49.6 | 13.0 | 40.6 | 10.7 |
| 70-89 | 67.2 | 29.6 | 62.8 | 28.4 |
| 90+ | 76.7 | 32.3 | 65.2 | 25.3 |
The mean recipient eGFR at 1 year adjusted for the covariates of donor age and donor-recipient age mismatch (“fit”), together with standard error (“se”) and 95% confidence intervals (“lower”, “upper”), is:
| GFR_group | fit | se | lower | upper |
|---|---|---|---|---|
| 60-69 | 67.9 | 8.5 | 51.2 | 84.7 |
| 70-89 | 70.4 | 2.4 | 65.6 | 75.1 |
| 90+ | 72.6 | 2.3 | 68.0 | 77.1 |
eGFR at 5 years
Df Sum Sq Mean Sq F value Pr(>F)
GFR_group 2 4771 2386 3.598 0.0305 *
Residuals 114 75585 663
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
263 observations deleted due to missingnessA multivariate model including donor age and donor-recipient age mismatch showed that the recipient eGFR at 5 year was not predicted by the donor GFR:
Anova Table (Type III tests)
Response: Rec_eGFR_5yr
Sum Sq Df F value Pr(>F)
(Intercept) 20516 1 35.0807 3.542e-08 ***
Age 10080 1 17.2358 6.466e-05 ***
Age_mismatch 2355 1 4.0267 0.04719 *
GFR_group 1096 2 0.9367 0.39496
Residuals 65499 112
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1Effect sizes:
term etasq
1 Age 0.143
2 Age_mismatch 0.029
3 GFR_group 0.014The mean recipient eGFR at 5 years adjusted for the covariates of donor age and donor-recipient age mismatch (“fit”), together with standard error (“se”) and 95% confidence intervals (“lower”, “upper”), is:
| GFR_group | fit | se | lower | upper |
|---|---|---|---|---|
| 60-69 | 57.7 | 9.1 | 39.7 | 75.7 |
| 70-89 | 66.3 | 3.8 | 58.8 | 73.8 |
| 90+ | 60.4 | 3.2 | 54.1 | 66.8 |
Recipient and transplant survival
Recipient survival
There were no significant differences in basic recipient survival based on donor iGFR group
There were no pre-transplant factors predicting post-transplant survival:
Call:
coxph(formula = recipient.surv ~ Rec_age + Age_mismatch + GFR_group,
data = df)
coef exp(coef) se(coef) z p
Rec_age 0.03212 1.03264 0.02861 1.123 0.262
Age_mismatch -0.01810 0.98206 0.02360 -0.767 0.443
GFR_group70-89 0.54804 1.72986 1.09800 0.499 0.618
GFR_group90+ 0.04985 1.05111 1.20250 0.041 0.967
Likelihood ratio test=10.21 on 4 df, p=0.03704
n= 321, number of events= 18
(59 observations deleted due to missingness)
Transplant survival
There were no significant differences in death-censored graft survival based on donor iGFR group
There were no pre-transplant factors predicting post-transplant survival:
Call:
coxph(formula = tx.surv ~ Rec_age + Age + Age_mismatch + Preop_sys_BP +
Preop_dia_BP + Latest_preop_ACR + GFR_group, data = df)
coef exp(coef) se(coef) z p
Rec_age -2.059e-02 9.796e-01 1.340e-02 -1.536 0.125
Age -4.036e-03 9.960e-01 2.420e-02 -0.167 0.868
Age_mismatch NA NA 0.000e+00 NA NA
Preop_sys_BP -3.144e-03 9.969e-01 1.512e-02 -0.208 0.835
Preop_dia_BP 7.299e-03 1.007e+00 2.933e-02 0.249 0.803
Latest_preop_ACR -5.480e-01 5.781e-01 4.882e-01 -1.123 0.262
GFR_group70-89 1.644e+01 1.374e+07 5.401e+03 0.003 0.998
GFR_group90+ 1.719e+01 2.933e+07 5.401e+03 0.003 0.997
Likelihood ratio test=10.02 on 7 df, p=0.1876
n= 290, number of events= 20
(90 observations deleted due to missingness)