This document summarizes a presentation on emerging factors for predicting adverse outcomes in kidney transplantation. The presentation discusses how prediction of outcomes is important for targeting interventions and precision medicine. Current clinical tools are limited in predicting long-term graft failure. Higher levels of proteinuria are an independent risk factor for graft failure. Proteinuria is a fair predictor of late graft failure, with its predictive accuracy improving over time after transplantation.

1. Emerging factors for predicting
adverse outcomes in kidney
transplantation
Maarten Naesens
52nd ERA-EDTA meeting
London, May 2015
Maarten Naesens, MD PhD
University of Leuven, Belgium

2. Prediction of adverse outcomes
is highly relevant
Therapeutic
targets
Timely &
targeted
intervention
Surrogate
endpoints
Precision medicine

3. The paradigm of precision medicine
PREDICT
Adapted from Atul Butte

4. PREDICT DIAGNOSE
PREVENT TREAT
WAIT
The paradigm of precision medicine

5. PREDICT DIAGNOSE
PREVENT TREAT
The paradigm of precision medicine

6. PREDICTING
GRAFT
FAILURE
Current
clinical
practice
Novel
tools
The way
to move
forward

7. Naesens M et al. Unpublished data
The future of a kidney transplant
is a lesson in humility
Death-censored graft survival
for patients transplanted at UZ Leuven
5 10 20 30
0
20
40
60
80
100
Time after transplantation (years)
Percentsurvival(%)
'80-'90 (N=747)
'90-'00 (N=1192)
'00-'05 (N=530)
'05-'10 (N=629)
Transplant era
'10-'15 (N=618)

8. Current clinical routine in kidney
transplantation uses an age-old toolbox
Graft
failure
Donor/recipient
demographics
Delayed
graft
function
Acute
rejection
Creatinine
eGFR
Proteinuria
Graft
histology

9. Delayed graft function is NOT associated with
worse outcome in DCD kidneys
From Summers et al. Lancet 2010
0.0
0.5
1.0
1.5
2.0
2.5
Hazardratio(95%CI)
forgraftfailure
p=0.29
p<0.001
DGF
No DGF
DBD
(25% DGF)
DCD
(50% DGF)
Hazard ratio for graft failure

10. Delayed graft function is NOT associated with
worse outcome in DCD kidneys
From Summers et al. Lancet 2010
0.0
0.5
1.0
1.5
2.0
2.5
Hazardratio(95%CI)
forgraftfailure
p=0.29
p<0.001
DGF
No DGF
DBD
(25% DGF)
DCD
(50% DGF)
Hazard ratio for graft failure

11. Budde et al Lancet 2011; Budde et al Am J Transplant 2014
eGFR or TCMR as surrogate endpoints
in renal transplantation?
5-year graft loss:
2.1% in CsA group
2.6% in EVR group
P = 1.00
Z ZEUS trial
rejection risk
(after switch)
3% in CsA group
10% in EVR group
P = 0.04

12. Rostaing, Vincenti et al Am J Transplant 2013
eGFR or TCMR as surrogate endpoints
in renal transplantation?
Belatacept LI
Belatacept MI
Cyclosporine
BENEFIT trial

13. BELA MI BELA LI CsA
0%
5%
10%
15%
14%
9%
6%
BELA MI BELA LI CsA
0
50
100
eGFR (mL/min/1.73m2)
BELA MI BELA LI CsA
0%
5%
10%
15%
20%
Acute rejection incidence
14%
9%
6%
BELA MI BELA LI CsA
0
50
100
eGFR (mL/min/1.73m2)
BELA MI BELA LI CsA
0%
50%
100%
Graft loss at 3 years
95% 96% 95%
BELA MI BELA LI CsA
0
50
100
IFTA grade > 0 at 1 year
19% 20%
44%
From Vincenti et al New Engl J Med 2005;Vincenti et al Am J Transplant 2010; Rostaing et al Am J Transplant 2013
The BENEFIT trial shows uncoupling of
acute rejection from eGFR and from failure
***
*
*** *
Higher rejection risk Better eGFR

14. Vincenti et al ATC May 2015 – Abstract #452
The BENEFIT trial shows uncoupling of
acute rejection from eGFR and from failure
Belatacept MI
Cyclosporine
Belatacept LI
PREDICT
Overall graft survival
Time after transplantation (months)
p<0.05

15. Not every risk factor is a good predictor.
A predictor needs to be accurate!

16. Accuracy of a diagnostic or predictive test
determines its clinical value, not its p-value!
Area under a
ROC curve
Interpretation
0.90 – 1.00 Excellent (A)
0.80 – 0.90 Good (B)
0.70 – 0.80 Fair (C)
0.60 – 0.70 Poor (D)
0.50 – 0.60 Fail (F)
False positive rate (1 – Specificity)
Truepositiverate(Sensitivity)
Perfect test
AUC=1.00
Good test
AUC=0.85
Failed test
AUC=0.50
PPV and NPV
take disease prevalence into account

17. Creatinine at 1 year is significantly associated
with graft outcome, but is a poor predictor
Kaplan et al Am J Transplant 2003
1 - Specificity
Sensitivity
AUC = 0.63
ROC curve:
Serum creatinine at 1 year as predictor for graft failure

18. ROC for graft failure
5 year after biopsy
according to 1 year MDRD eGFR
0 20 40 60 80 100
0
20
40
60
80
100
100% - Specificity%
Sensitivity%
AUC=0.77
p<0.0001
eGFR at 1 year is significantly associated with
graft outcome, but is only a fair predictor
MRDR eGFR at 1 year
and graft failure
1 5 10 15
0
20
40
60
80
100
Time after biopsy (years)
Graftsurvival(%)
>70 mL/min
60-70 mL/min
50-60 mL/min
log-rank
P<0.0001
40-50 mL/min
30-40 mL/min
20-30 mL/min
<20 mL/min
Naesens et al (Unpublished)

19. Proteinuria is an independent risk factor of
kidney graft failure
Naesens M et al J Am Soc Nephrol – in press
3 months
(N=914)
1 5 10 15
0
20
40
60
80
100
Time after transplantation (years)
Percentsurvival
>3.0 g/24h
< 0.3 g/24h
0.3-1.0 g/24h
1.0-3.0 g/24h
log-rank
P <0.0001
No. at risk
<0.3 g/24h
0.3-1.0 g/24h
1.0-3.0 g/24h
>3.0 g/24h
733
136
37
8
720
125
35
8
636
107
29
5
497
76
19
2
173
28
11
1
Proteinuria
2 years
(N=731)
5 10 152
0
20
40
60
80
100
Time after transplantation (years)
Percentsurvival
< 0.3 g/24h
0.3-1.0 g/24h
> 1.0 g/24h
572
119
430
68
163
23
532
102
Proteinuria
log-rank
P <0.0001
No. at risk
<0.3 g/24h
0.3-1.0 g/24h
1 year
(N=778)
1 5 10 15
0
20
40
60
80
100
Time after transplantation (years)
Percentsurvival
< 0.3 g/24h
0.3-1.0 g/24h
> 1.0 g/24h
614
123
41
453
66
19
166
20
8
561
93
30
Proteinuria
log-rank
P <0.0001
No. at risk
<0.3 g/24h
0.3-1.0 g/24h
>1.0 g/24h
5 years
(N=637)
5 10 15
0
20
40
60
80
100
Time after transplantation (years)
Percentsurvival < 0.3 g/24h
0.3-1.0 g/24h
> 1.0 g/24h
495
104
160
28
416
70
Proteinuria
log-rank
P <0.0001
No. at risk
<0.3 g/24h
0.3-1.0 g/24h
A
3 months
(N=914)
1 5 10 15
0
20
40
60
80
100
Time after transplantation (years)
Percentsurvival
>3.0 g/24h
< 0.3 g/24h
0.3-1.0 g/24h
1.0-3.0 g/24h
log-rank
P <0.0001
No. at risk
<0.3 g/24h
0.3-1.0 g/24h
1.0-3.0 g/24h
>3.0 g/24h
733
136
37
8
720
125
35
8
636
107
29
5
497
76
19
2
173
28
11
1
Proteinuria
2 years
(N=731)
5 10 152
0
20
40
60
80
100
Time after transplantation (years)
Percentsurvival
< 0.3 g/24h
0.3-1.0 g/24h
> 1.0 g/24h
572
119
40
430
68
16
163
23
7
532
102
33
Proteinuria
log-rank
P <0.0001
No. at risk
<0.3 g/24h
0.3-1.0 g/24h
>1.0 g/24h
1 year
(N=778)
1 5 10 15
0
20
40
60
80
100
Time after transplantation (years)
Percentsurvival
< 0.3 g/24h
0.3-1.0 g/24h
> 1.0 g/24h
614
123
41
453
66
19
166
20
8
561
93
30
Proteinuria
log-rank
P <0.0001
No. at risk
<0.3 g/24h
0.3-1.0 g/24h
>1.0 g/24h
5 years
(N=637)
5 10 15
0
20
40
60
80
100
Time after transplantation (years)
Percentsurvival
< 0.3 g/24h
0.3-1.0 g/24h
> 1.0 g/24h
495
104
38
160
28
7
416
70
13
Proteinuria
log-rank
P <0.0001
No. at risk
<0.3 g/24h
0.3-1.0 g/24h
>1.0 g/24h
A

20. Proteinuria is a fair predictor of graft failure
LATE after transplantation
Naesens M et al J Am Soc Nephrol (In press)
3 months
(N=914)
0 20 40 60 80 100
0
20
40
60
80
100
False Positive Fraction (%)
TruePositiveFraction(%)
AUC=0.64
(95% CI 0.57-0.70)
p<0.0001
1 year
(N=778)
0 20 40 60 80 100
0
20
40
60
80
100
False Positive Fraction (%)
TruePositiveFraction(%)
AUC=0.73
(95% CI 0.66-0.80)
p<0.0001
2 years
(N=731)
0 20 40 60 80 100
0
20
40
60
80
100
False Positive Fraction (%)
TruePositiveFraction(%)
AUC=0.71
(95% CI 0.63-0.80)
p<0.0001
5 years
(N=637)
0 20 40 60 80 100
0
20
40
60
80
100
False Positive Fraction (%)
TruePositiveFraction(%)
AUC=0.77
(95% CI 0.70-0.83)
p<0.0001
Glomerular disease
(N=86)
1 5 100
0
20
40
60
80
100
Time after biopsy (years)
Percentsurvival
Proteinuria >3.0 g/24 hours
Proteinuria < 0.3 g/24 hours
Proteinuria 0.3-1.0 g/24 hours
Proteinuria 1.0-3.0 g/24 hours
log-rank
P < 0.0001
25
17
27
15
21
13
13
6
17
7
5
3
9
4
1
1
No. at risk
<0.3 g/24h
0.3-1.0 g/24h
1.0-3.0 g/24h
>3.0 g/24h
Transplant glomerulopathy
(N=101)
1 5 100
0
20
40
60
80
100
Time after biopsy (years)
Percentsurvival
Proteinuria >3.0 g/24 hours
Proteinuria < 0.3 g/24 hours
Proteinuria 0.3-1.0 g/24 hours
Proteinuria 1.0-3.0 g/24 hours
log-rank
P = 0.0005
14
28
42
17
10
19
22
4
7
6
8
0
3
1
1
0
No. at risk
<0.3 g/24h
0.3-1.0 g/24h
1.0-3.0 g/24h
>3.0 g/24h
D
E Glomerular disease
(N=84)
1 5 100
0
20
40
60
80
100
Time after biopsy (years)
Percentsurvival
Proteinuria >3.0 g/24 hours
Proteinuria < 0.3 g/24 hours
Proteinuria 0.3-1.0 g/24 hours
Proteinuria 1.0-3.0 g/24 hours
log-rank
P < 0.0001
25
17
27
15
21
13
13
6
17
7
5
3
9
4
1
1
No. at risk
<0.3 g/24h
0.3-1.0 g/24h
1.0-3.0 g/24h
>3.0 g/24h
1 5 10
0
20
Time after biopsy (years)
P
No. at risk
<0.3 g/24h
0.3-1.0 g/24h
1.0-3.0 g/24h
>3.0 g/24h
548
319
150
30
449
237
94
18
297
156
53
11
646
390
208
51
log-rank
P <0.0001
0 20 40 60 80 100
0
20
False Positive Fraction (%)
TruePo
AUC=0.66
(95% CI 0.63-0.69)
P <0.0001
Time posttransplant
AUC=0.64
p<0.0001
AUC=0.73
p<0.0001
AUC=0.71
p<0.0001
AUC=0.77
p<0.0001
Proteinuria (>1g/24h) as marker for graft failure at 5 years after measurement
Time point Sensitivity Specificity PPV NPV
3 months 10.1% (4.47%-19.0%) 95.3% (93.7%-96.7%) 19.5% 91.2%
1 year 16.4% (8.15%-28.1%) 95.5% (93.8%-96.9%) 26.3% 92.6%
2 years 20.0% (10.4%-33.0%) 95.6% (93.7%-97.0%) 30.6% 93.1%
5 years 28.4% (18.0%-40.7%) 96.4% (94.5%-97.7%) 61.3% 91.0%

21. Baseline biopsy histology is a
good predictor of graft failure
Leuven Donor Risk Score
= GS + 3xIFTA + donor age
Graft failure at 5 year
post-transplant
AUC = 0.81
P<0.0001
P<0.0001
De Vusser, Naesens et al J Am Soc Nephrol 2013

22. Despite significant association with failure,
the CADI score is less accurate as predictor
Naesens et al (Unpublished)
CADI score
in indication biopsy
1 5 10 15
0
20
40
60
80
100
Time after biopsy (years)
Graftsurvival(%)
CADI 0
CADI 1
CADI 2-3
log-rank
P<0.0001
CADI 4-5
CADI 6-7
CADI 8-9
CADI >9
ROC for graft failure
5 year after biopsy
according to CADI score
0 20 40 60 80 100
0
20
40
60
80
100
100% - Specificity%
Sensitivity%
AUC=0.65
p<0.0001
CADI score
in indication biopsy
1 5 10 15
0
20
40
60
80
100
Time after biopsy (years)
Graftsurvival(%)
CADI 0
CADI 1
CADI 2-3
log-rank
P<0.0001
CADI 4-5
CADI 6-7
CADI 8-9
CADI >9
ROC for graft failure
5 year after biopsy
according to CADI score
0 20 40 60 80 100
0
20
40
60
80
100
100% - Specificity%
Sensitivity%
AUC=0.76
p<0.0001
Late biopsies:
ROC for 5y graft loss
(>1y)
N=1335 indication biopsies

23. PREDICTING
GRAFT
FAILURE
Current
clinical
practice
Novel
tools
The way
to move
forward

24. Increased urinary KIM-1 mRNA expression
associates with graft injury and failure
Szeto et al CJASN 2010
Low KIM-1
High KIM-1
p=0.006
Histology Function Survival

25. Urinary KIM-1 mRNA expression is a
poor predictor of graft failure
Szeto et al CJASN 2010
AUC=0.68
P=0.02
Best cut-off value for urinary KIM-1
for graft failure:
- sensitivity 68%
- specificity 68%
- PPV 65%
- NPV 68%
ROC AUC 0.68

26. Urinary CCL2:creatinine ELISA is a
poor predictor of graft failure
Ho, Nickerson, Rush et al et al Transplantation 2013
AUC=0.68
P=0.02
Best cut-off value for urinary
CCL2 ELISA for graft failure:
- Sensitivity 70%
- Specificity 70%
- PPV 19%
- NPV 96%
Death-censored graft survival
Time after transplantation (years)

27. Increased plasma sCD30 at day 30 postTX
associates with graft failure
Susal, Opelz et al Transplantation 2011
Low sCD30
High sCD30
Best cut-off value for
plasma sCD30
for graft failure:
- Sensitivity 18%
- Specificity 92%
- PPV 22%
- NPV 90.3%

28. “Novel” urinary markers for graft failure
are less predictive than albuminuria
Nauta, Gansevoort et al Am J Kidney Dis 2011
AUC = 0.67
AUC = 0.78
AUC = 0.74
AUC = 0.75
AUC = 0.62
AUC = 0.63
FAIR IS NOT GOOD ENOUGH

29. Donor-specific antibodies associate with graft
failure, but are not a crystal ball
Wiebe, Rush et al Am J Transplant 2012
No de novo DSA
de novo DSA
Sensitivity = 64%
Specificity = 89%
PPV = 29.7%
NPV = 97.0%
Death-censored graft survival
Time after transplantation (years)

30. Molecular “ABMR score” predicts graft failure
better than histology of ABMR
INTERCOM STUDY
(multicenter)
ABMR Score -
Histology -
ABMR Score -
Histology +
ABMR Score +
Histology +
ABMR Score +
Histology -
Halloran et al Am J Transplant 2013
ABMR score for graft loss:
Sensitivity = 75%
Specificity = 81%
PPV = 48%
NPV = 93%
ROC AUC=0.81

31. Molecular “Risk score” predicts graft outcome
better than histology or proteinuria
Low risk score
High risk score
Time after biopsy
Survivalprobability
Einecke et al J Clin Invest 2010
AUC=0.83
Risk score for graft loss:
Early biopsies:
Sensitivity = 100%
Specificity = 41%
PPV = 5%
NPV = 100%
Late biopsies:
Sensitivity = 83%
Specificity = 63%
PPV = 47%
NPV = 90%

32. Invasive and noninvasive markers of graft loss
have moderate predictive performance
Kaplan et al AJT 2003; Szeto et al CJASN 2010; Einecke et al JCI 2010; Susal et al Transplantation 2011;
Wiebe et al AJT 2012; Ho et al Transplantation 2013; Halloran et al Am J Transplant 2013; Naesens et al JASN 2015;
0.0 0.2 0.4 0.6 0.8 1.0
0.0
0.2
0.4
0.6
0.8
1.0
PPV
NPV
urinaryKIM-1
Molecular RiskScore(early)
Molecular RiskScore (late)
urinary CCL2:creatinine
plasmasCD30
24h-proteinuriaat 5yde novo DSA
ABMR score
serum creatinineat 1year

33. Invasive and noninvasive markers of graft loss
have moderate predictive performance
Kaplan et al AJT 2003; Szeto et al CJASN 2010; Einecke et al JCI 2010; Susal et al Transplantation 2011;
Wiebe et al AJT 2012; Ho et al Transplantation 2013; Halloran et al Am J Transplant 2013; Naesens et al JASN 2015;
0.0 0.2 0.4 0.6 0.8 1.0
0.0
0.2
0.4
0.6
0.8
1.0
PPV
NPV
urinaryKIM-1
Molecular RiskScore(early)
Molecular RiskScore (late)
urinary CCL2:creatinine
plasmasCD30
24h-proteinuriaat 5yde novo DSA
ABMR score
serum creatinineat 1year
0.0 0.2 0.4 0.6 0.8 1.0
0.0
0.2
0.4
0.6
0.8
1.0
PPV
NPV
urinaryKIM-1
Molecular RiskScore(early)
Molecular RiskScore (late)
urinary CCL2:creatinine
plasmasCD30
24h-proteinuriaat 5yde novo DSA
ABMR score
serum creatinineat 1year
0.0 0.2 0.4 0.6 0.8 1.0
0.0
0.2
0.4
0.6
0.8
1.0
PPV
NPV
urinaryKIM-1
Molecular RiskScore(early)
Molecular Risk Score(late)
urinary CCL2:creatinine
plasmasCD30
24h-proteinuriaat 5yde novo DSA
ABMR score
serum creatinineat 1year
NPV > 90%
High NPV

34. Invasive and noninvasive markers of graft loss
have moderate predictive performance
Kaplan et al AJT 2003; Szeto et al CJASN 2010; Einecke et al JCI 2010; Susal et al Transplantation 2011;
Wiebe et al AJT 2012; Ho et al Transplantation 2013; Halloran et al Am J Transplant 2013; Naesens et al JASN 2015
0.0 0.2 0.4 0.6 0.8 1.0
0.0
0.2
0.4
0.6
0.8
1.0
PPV
NPV
urinaryKIM-1
Molecular RiskScore(early)
Molecular RiskScore (late)
urinary CCL2:creatinine
plasmasCD30
24h-proteinuriaat 5yde novo DSA
ABMR score
serum creatinineat 1year
0.0 0.2 0.4 0.6 0.8 1.0
0.0
0.2
0.4
0.6
0.8
1.0
PPV
NPV
urinaryKIM-1
Molecular RiskScore(early)
Molecular RiskScore (late)
urinary CCL2:creatinine
plasmasCD30
24h-proteinuriaat 5yde novo DSA
ABMR score
serum creatinineat 1year
0.0 0.2 0.4 0.6 0.8 1.0
0.0
0.2
0.4
0.6
0.8
1.0
PPV
NPV
urinaryKIM-1
Molecular RiskScore(early)
Molecular Risk Score(late)
urinary CCL2:creatinine
plasmasCD30
24h-proteinuriaat 5yde novo DSA
ABMR score
serum creatinineat 1year
Low PPV
0.0 0.2 0.4 0.6 0.8 1.0
0.0
0.2
0.4
0.6
0.8
1.0
PPV
NPV
urinaryKIM-1
Molecular RiskScore(early)
Molecular RiskScore(late)
urinaryCCL2:creatinine
plasmasCD30
24h-proteinuriaat 5yde novo DSA
ABMR score
serum creatinineat 1year
PPV < 60%
Low PPV

35. PREDICTING
GRAFT
FAILURE
Current
clinical
practice
Novel
tools
The way
to move
forward

36. Suggested pathways to move things forward
Combined
markers
Collaboration
External validation
Population
specificity
Statistical
rigour

37. British Government, 1939