Transcript sept28_2006Discussion

Case Studies in Modeling and
Simulation
Discussion
Stella G. Machado, Ph.D.
Office of Biostatistics/OTS/CDER/FDA
FDA/Industry Workshop, September 2006
1
Regulatory issue
• Approval was sought for monotherapy
for pediatric population, without another
clinical trial
• Clinical trial data for Drug X:
– Adults: adjunct and monotherapy
– Pediatric population: adjunct
• PK/PD modeling used for bridging the
adjunct therapy data (data masked)
2
Bridging PK/PD Studies
• General method comparing PK/PD response curves in:
Pediatric versus Adult populations
Different Regions
• Exposure: dose, AUC, Cmin, etc
• Response: biomarkers, clinical endpoints
• Goal is to evaluate similarity in PK/PD relationships
between 2 populations
Conclude: similarity, similarity with some dose regimen
modification; lack of similarity
3
DRUG X: PK/PD scatter plot with loess fits
5
0
N
e
w
O
r
i
g
i
n
a
l
4
0
Respon
3
0
2
0
1
0
0
0
2
0 4
0 6
0 8
0 1
0
01
2
0
C
o
n
c
.
4
STEPS IN THE STATISTICAL
APPROACH
• assess similarity between responses at all
concentrations likely to be encountered
• account for variability of the response
• need “Equivalence” type approach, not hypothesis
tests showing that the responses are not significantly
different
• analysis is more “exploratory” than “confirmatory”
5
Steps
• Usual equivalence-type analysis:
– “similarity” defined as requirement that average responses in
the 2 populations, at the same C, are closely similar:
– choose reference “goalposts” L and U, eg 80% to 125%
– calculate 95% confidence interval for ratio of average
responses (1 / 0) for “all” C
6
EXAMPLE: Drug X
• Response transformed by square root to stabilize the
variance
• Linear models fitted separately for the two
populations:
• sqrt(response) = a + b * Conc + 
• For each C, 5000 pairs of studies generated  5000
estimates of 1/0, and percentiles
7
DRUG X: 95% CI’s for ratios 1/0 for
concentrations: 0, 20,50,70,90 via
model-based method
1
.
8
95%confidebusorimlaty
1
.
4
1
.
0
0
.
6
0
4
0
8
0
1
2
0
C
o
n
c
8
Remarks for example
• Response higher for pediatric population for
concentrations above 50mg
=>Shows lack of similarity, but dose adjustment
would be possible if high concentrations are called
for
• Limits of (80, 125) might not be medically most
sensible for interpretation in each situation
9