40
CPA (%)
A SEMI-MECHANISTIC MODEL OF FIBROSIS PROGRESSION USING COLLAGEN PROPORTIONATE AREA
(CPA) IN POST LIVER TRANSPLANT (POLT) PATIENTS WITH CHRONIC HEPATITIS C (CHC)
30
20
10
BI=cyclosporine
AHC=NO
BI: cyclosporine
AHC=YES
50
40
30
G.
1
Ma ,
A.
3
Dhillon ,
A.
3
Hall ,
A.K.
2
Burroughs ,
E.
2
Tsochatzis ,
G.
4
Burgess ,
F.
20
1
Guo
10
0
5
10
15
1Clin Pharm&PMx, Pfizer, China; 2Royal Free Sheila Sherlock Liver Centre and 3Univ Dept of Histopathology, Royal Free Hospital, UK; 4Conatus Pharmaceuticals, USA
Time (year)
Observed CPA by AHC
0
5
AHC=NO
Observed CPA by Basic Immunosuppression
10
15
0
AHC=YES
Observed CPA by AHC+BI
50
15
BI=tacrolimus
AHC=YES
20
30
20
10
10
0
0
5
10
50
15
0
Observed CPA by AHC+BI
Collagen
10
40
5
15
Generation
Degradation
BI=tacrolimus
AHC=NO
Kout*A
20
2
10
0
5
10
5
15
0
Observed CPA by AHC
5
10
10
Observed CPA by AHC
15
0
BI=cyclosporine
0
AHC=YES
AHC=NO
5
5
10
Time (year)
10
Observed CPA by Basic Immunosuppression
15
BI=tacrolimus15
0
AHC=YES
50
CPA (%)
30
40
5
BI=cyclosporine
40
50
15
Time (year)
Observed CPA by Basic Immunosuppression
5
50
10
BI=tacrolimus
50
30
40
0
Collagen generation is at a constant rate.
No collagen degradation component.
5
10
10
30
0
20
15
0
5
10
Time (year)
10
Collagen generation is associated with collagen proportion.
No collagen degradation component.
20
15
Time (year)
30
10
Predicted CPA by AHC+BI
0
BI=tacrolimus
AHC=NO 0
5
10
5
10
15
0
BI=tacrolimus
AHC=YES
15
0
Time (year)
5
10
15
Time (year)
50
Figure. 5 Simulation Plots
40
% Fibrosis (CPA) Progression (2000
Simulations)
Predicted
CPA by AHC+BI
20
ID:19
ID:21
ID:24
ID:27
ID:28
ID:29
ID:30
ID:31
20
15
10
5
0
ID:33
ID:37
ID:40
ID:42
ID:45
ID:47
ID:49
15
Time (year)
BI=cyclosporine
AHC=NO
2
3
4
5
6
ID:57
ID:59
ID:60
ID:61
ID:62
0
5 10 15
5
0 5 10 15
Time (year)
100
Overall Patients after Biopsy
>9
0
1
2
3
4
5
6
7
Time (year)
8
9
>9
Time (year)
10
30<=CPA
28<=CPA<30
26<=CPA<28
24<=CPA<26
22<=CPA<24
20<=CPA<22
18<=CPA<20
16<=CPA<18
14<=CPA<16
12<=CPA<14
10<=CPA<12
8<=CPA<10
6<=CPA<8
4<=CPA<6
2<=CPA<4
0<=CPA<2
15
0
20
RESULTS
where Kin is generation rate, Kout is degradation rate, AHC =
acute HCV status, and is a binary variable (0=NO; 1=YES), BI =
basic immunosuppression, and is binary (0=cyclosporine,
1=tacrolimus).
Estimated Kin and Kout were 0.019 and 0.48, respectively. ISV
was 67% and 92% for Kin and Kout, respectively. Patients with
tacrolimus treatment appeared to have their collagen
generation rate increased by 0.027 compared to those with
cyclosporine treatment. Patients with acute HCV appeared to
have their collagen generation rate increased by 0.016
compared to those without acute HCV.
9
Time (year)
% Simulations
0
8
% Fibrosis (CPA) Progression (2000 Simulations)
20
15
10
5
0
0 5 10 15
7
60
ID:56
80
0
AHC=NO
10
ID:53
60
50
30
0
1
20
ID:50
10
BI=cyclosporine
AHC=YES
AHC=YES
40
20
15
10
5
0
ID:51
40
10
40
CPA (%)
5
20
20
0
30
0
ID:17
40
% Fibrosis (CPA) Progression (2000 Simulations)
100
0
1
2
3
4
5
6
% Fibrosis (CPA) Progression (2000 Simu
7
8
9
>9
20
40
60
80
Time (year)
30<=CPA
28<=CPA<30
26<=CPA<28
24<=CPA<26
22<=CPA<24
20<=CPA<22
18<=CPA<20
16<=CPA<18
14<=CPA<16
12<=CPA<14
10<=CPA<12
8<=CPA<10
6<=CPA<8
4<=CPA<6
2<=CPA<4
0<=CPA<2
100
ID:16
20
15
10
5
0
50
80
ID:11
10
BI=tacrolimus
AHC=YES
60
ID:9
0 5 10 15
15
40
ID:7
5 10 15
30<=CPA
28<=CPA<30
26<=CPA<28
24<=CPA<26
22<=CPA<24
20<=CPA<22
18<=CPA<20
16<=CPA<18
14<=CPA<16
12<=CPA<14
10<=CPA<12
8<=CPA<10
6<=CPA<8
4<=CPA<6
2<=CPA<4
0<=CPA<2
10
20
ID:6
0
5
% Simulations
ID:4
5 10 15
100
ID:2
0
20
80
0 5 10 15
Population Prediction
0
% Simulations
100
80
40
Predicted CPA (%)
Individual Prediction
AHC=YES
50
30
Observed
10
BI=tacrolimus
AHC=NO
BI=cyclosporine
60
Figure. 3 Examples of Individual Predictions /Goodness of Fit Plot
% Fibrosis (CPA) Progression (2000 Simu
30
20
BI=cyclosporine
AHC=NO
40
2841.11
% Simulations
Collagen generation is associated with normal liver proportion,
while collagen degradation is associated with collagen proportion.
Predicted CPA (%)
 3
dA
 kin  (100  A)  kout  A
dt
15
20
% Simulations
Basic Immunosuppression= Cyclosporine
Basic Immunosuppression= Tacrolimus
0
0
50
40
30
20
10
0
10
15
Time (year)
10
4384.96
40
10
CPA (%)
dA
 kin  A
dt
50
20
20
2848.56
BI=cyclosporine
AHC=YES
30
20
OFV
dA
 kin
dt
BI=cyclosporine
AHC=NO
10
40
Tested Structures:
30
10
30
: Collagen Proportion
40
20
BI: cyclosporine
AHC=YES
30
AHC=NO
A
30
Predicted CPA (%)
40
BI=cyclosporine
AHC=NO
40
dCPA
 (kin  AHC  BI )  (100  CPA)  kout  CPA
dt
Pharmacometrics,
Global Clinical Pharmacology
BI=tacrolimus
AHC=YES
10
BI: cyclosporine
AHC=YES
10
50
The analysis showed that acute HCV, basic immunosuppression,
and number of immunosuppressive regimens impacted collagen
generation rate while donor age was correlated with collagen
degradation rate. After covariate forward inclusion and
backward exclusion steps, the following model best fitted
patients' CPA progression data:
Time after LT (years)
15
20
0
0 5 10 15
15
10
20
50
40
BI=cyclosporine
AHC=NO
0
10
5
BI=tacrolimus
AHC=NO
CPA (%)
Kin*(100-A)
100-A: Normal Liver Proportion
1
Figure. 1 CPA Raw Data from 185 consecutive POLT CHC patients
0
50
0
Longitudinal CPA data together with demography, HCV
genotype, medical and immunosuppressive therapy from 185
consecutive POLT CHC patients [Figure. 1] were included in the
model. Assumptions regarding generation and degradation
mechanisms for collagen [Figure. 2] were made for CPA
modeling. Fixed effects and inter-subject variability (ISV) were
estimated by a nonlinear mixed effects modeling technique.
Acute HCV, basic immunosuppression, and donor age were
potentially associated with CPA progression. The covariates in
the model were selected by a stepwise forward inclusion (p<
0.05) and backward (p< 0.005) exclusion method.
The model was evaluated using graphical and numerical
diagnostic tools [Figure. 3].
15
BI=tacrolimus
AHC=YES
CPA (%)
Degradation Rate
CPA (%)
Generation Rate
CPA (%)
METHODS
10
50
0
To build a semi-mechanistic model of progression of CPA for
POLT CHC patients and explore potential patient characteristics
affecting quantity of liver fibrosis.
5
Time (year)
30
Predicted CPA by AHC+BI
0
15
40
CPA (%)
10
Time (year)
AIM
CPA (%)
30
0
dA
 kin  (100  A)  kout  A
dt
5
5
Figure. 4 Observed and Predicted CPA Stratified by Basic
Immunosuppression and Acute HCV status
Figure. 2 Base Model Structure Interpretation
Quantitative measurements of collagen proportionate area
(CPA) by digital image analysis quantifies fibrosis in the liver. A
semi-mechanistic model was developed to describe CPA
progression in POLT CHC patients.
0
CPA (%)
BACKGROUND
BI=tacrolimus
AHC=NO
10
BI=tacrolimus
50
0
40
5
BI=cyclosporine
0
1
2
3
4
5
6
7
8
9
>9
Time (year)
0
1
2
3
4
5
6
7
8
9
>9
Time (year)
CONCLUSIONS
A mechanism-based methodology could be applied to model
CPA progression. A semi-mechanistic CPA progression model
developed from POLT CHC patients suggests that use of
tacrolimus and acute HCV infection significantly increase the
rate of collagen generation [Figure. 4 and Figure. 5]. The CPA
model may evolve with further investigation of mechanisms of
collagen generation and degradation and exploration of
variations of immunosuppression.