Transcript HilaDavid-ShimritVashdi

Verify or refute
the use of Non Linear Mixed
Effect Model for Interferon effect
on HCV
Hila David
Shimrit Vashdi
Project Advisors:
Prof. Avidan Neumann
Dr. Rachel Levy Drummer
Introduction
Biomathematical Model is a valuable tool for
science and it has implications on medicine and
economy.
It is often used to characterize diseases and
drug’s behavior at the human body.
Finding the right model for HCV
treatment will have a great medical and
economic influence.
Hepatitis C Virus
HCV is a Single
Strand Ribonucleic
acid (RNA), belongs
to the Flaviviridae
family.
Its genome is 9.6 kb
size, and encoding to
a polyprotein of 3,000
amino acids,
produced by cellular
and viral proteases.
Interferon α
IFN- α is an anti-viral treatment for HCV. It’s a
Glycoprotein, naturally secreted from cells in
a response to viral infection.
The Glycoprotein attach to membrane
receptors which starts a cellular signals
sequence. Those signals cause expression
of anti-viral genes.
Pegylated-Interferon α
Polyethylen glycol (peg) is a polymer which
improves the pharmacokinetiks &
pharmacodynamics of proteins its attached to.
Two variants of pegylated-IFN α were tested,
pegasys and pegIntron, differ each other with
three features which effect their behavior:
• average molecular weight.
• branching.
• Link to the Interferon.
PharmacoKinetics
Study of the absorption, spreading, metabolism and elimination of a
drug.
Its important to understand the IFN-α pharmacokinetics in order to
efficiently predict the patients response to the treatment, since it’s a
critical stage of the disease.
The equations describes the concentration of the drug as a function of
time.
The first relates to the bolus and the second to the serum.
Bifn- the drug concentration at the bolus.
Sifn- drug concentration at the serum.
Inj- the drug dose.
Cifn- drug elimination rate.
Kbs- spreading drug rate.
Project Goal
Running a simulation with virtual
patients, using Non-Linear Mixed
Effect Model in order to verify or
refute the use of the individual
model for IFN-α effect on HCV.
Individual PK
The data is blood samples collected for each
patient separately and the estimation of the
parameters is done for each patient
specifically.
Attributes:
• Independency of the patients.
• More complicated to implement.
Population PK
Estimation of population parameters by
treating all data as if it arose from homogeneous
population. It can also identify the sources of
variability that explain differences in the
parameters between patients.
Attributes:
• More objective.
• Easier to implement.
• More powerful (under some assumptions).
Non Linear Mixed Effect Model
for PK
A method based on population PK.
NLME makes a one stage analysis and evaluate
the population parameters that enable determine
the PK and PD simultaneously.
The NLME combine both approaches, the
individual and population PK.
It fits the best model under statistic population
assumptions and can combine together
parameters with different influence.
MONOLIX PROGRAM
Monolix is a new software for the analysis of
Non-linear mixed effect models, used
especially at clinical experiments and
pharmacokinetics processes.
Monolix requires to define the data and the
model and to fix some parameters used for
the algorithms.
The output is the estimation of the individual
parameters, the maximal likelihood and the
residuals.
Working process
Analysis of Individual Experimental Data
• Kinetics graphs.
• Individual parameters.
Creating data for virtual patients
• Simulated Individual kinetic profiles.
• Adding noise to the simulated Individual kinetic profiles.
Running the population approach NLME
• Individual parameters out of population parameters.
Comparison of the methods
• Comparing the two methods individual parameters results.
*The working process was done for each treatment group of patients.
Step 1 – kinetics graphs
The drug concentration was measured
during the first week of the treatment at 21
patients treated with pegasys and 10
patients treated with pegintron.
• pegIntron
• pegasys
Pegasys Kinetics
PegIntron kinetics
3.5
concentration (log)
2.5
2
PegIntron kinetics
1.5
1
0.5
concentration (log pg/ml)
4.5
3
4
3.5
3
2.5
2
1.5
1
0.5
0
0
0.0
2.0
4.0
6.0
8.0
10.0
12.0
14.0
16.0
0
2
4
6
8
tim e(days)
time (days)
10
12
14
16
Step 2 - individual parameters
Running the real data with the model
equations at the Madonna.
Finding the combination of the parameters
values that will make the best fit of the real
data to the model for each patient.
pegIntron
pegasys
Step 3- creating virtual patients
• Creating 100 combinations of parameters
for each treatment.
• Simulating the kinetic profiles according to
the parameters of the individual patients.
• Adding noise (uniform distribution) on the
data outcomes from the kinetic profiles.
Step 4 - virtual patient’s
individual fit
Running the virtual patients data at the
Madonna and finding the individual fit and
parameters to every patient.
pegIntron
pegasys
Virtual parameters according to the
individual approach
pegasys
pegIntron
Cifn
Kbs
inj
Mean
0.499
0.498
s.d.
0.328
Cifn
Kbs
inj
83,280.4 11.711
0.423
51,506.6
0.261
54,875.2 7.1912
0.186
34,591.8
Minimum
1.367E-7 0.088
8,089.73 1.4614
0.01838
9,850.33
maximum
1.362
1.301
240,315
0.8315
293,502
median
0.438
0.431
67,146.7 11.4427
0.435
51,172.2
57.6864
Cifn Histogram
pegasys
pegIntron
Inj Histogram
pegasys
pegIntron
Kbs Histogram
Pegasys
pegIntron
Step 5 – population fit
Running the simulated data in monolix
program in order to estimate the population
parameters and the outcomes individual
parameters
Individual fitIndividual approach vs. population
approach
pegIntron
Blue- individual
Pink- population
pegasys
Red- individual approach
Blue- population approach
conclusions
• At the dynamic model, we can see clear
differences at Cifn and Inj between the
treatments, while the absorption from the bolus
to the serum (Kbs) is similar.
• Under the restriction of running the programs for
only one injection and for limited number of
patients, the model used at the monolix succeed
predicting the individual fits, but still the
individual approach find a better fit.
Thanks
• Prof. Avidan Neumann
• Dr. Rachel Levy Drummer