Hello everyone,Back to the Top
How should a phase-I trial to assess dose-proportionality in five
different doses be arranged? Where can a find a model for this trial ?
How can a PK/PD trial be included in it?
n: 54-11-4379-4200 int. 4099
Valeria,Back to the Top
There are a couple of questions you need to ask here. How long is the
halflife of the drug. Is it so long that a cross-over study design is
prohibitive. If not, then that is how you want to run this trial - as a
randomized crossover. If not, then you need a parallel group design.
As for a model, do you mean a model for the experimental design, like
sample publications or do you mean model in the sense of how do you
analyze the data. If you mean the latter, the best way is usually a
power model (See Gough et al., Drug Information Journal 29; 1995;
1039-1048) using the modification of Smith et al. (Pharmaceutical
Research 17; 2000; 1278-1283). As for whether you need pk-pd, pd is not
usually part of the trial but I don't see why it couldn't be. The data
would just be more difficult to analyze but may be more informative if
you have a saturation of pd effect.
Dear Valeria,Back to the Top
If you mean how a design of such a study should look like, than I can
give you information on a design that is often used as first dose in
man study (single dose escalation).
A relatively standard design for a SD escalation study is to study 6
increasing dose levels in two alternating panels of 9 subjects (total n
= 18). Thus each subject is studied in three times.
At each dose level 6 subjects will receive active drug substance and 3
subjects will receive placebo.
Dose-propotionality can be assessed graphically or statistically.
Graphically can be best done by making dose-normalized plots of the
relevant PK parameters. The overall mean of the dose-normalized
parameter can be plotted as horizontal line in de plot. Deviations
below or above the line indicate deviations from dose-proportionality.
Some statistical evaluations have already be mentioned by Peter Bonate.
I would like to add a linear mixed effect modeling procedure on the
dose-normalized parameters in which you have dose as fixed effect. With
contrasts you can make pair-wise comparisons between the different dose
levels and explore statistical significant differences.
The placebo subjects allow you to collect baseline information on the
incidence of AEs (if you like to do PK/PD on that) or on the
variability of certain pharmacodynamic measures/biomarkers of interest.
Kees Bol, PhD
Director Clinical Research
Consultant in Drug Development
Lage Mosten 29
4822 NK Breda
tel: +31 (0) 76 54 80 621
fax: +31 (0) 76 54 21 777
Nice post by Kees. The plot I like for evaluation of linearity of aBack to the Top
relevant parameter (e.g., clearance) is as follows (for clearance, but
it can be modified for Cmax, etc.):
CL(mL/min/kg) vs. Dose (mg/kg). You can then test the slope to see if
it is statistically significantly different from zero. Also, visual
inspection simply illustrates non-linearities, even those which don't
achieve statistical significance. For example, if the plot slopes
downward, this would indicate CL is reduced as dose increases. One can
also do the fancy statistical tests, but this is simple and reliable,
and you don't need to be a statistician to apply or understand it. Of
course, I expect the "hard-core" statisticans may wish to apply more
vigorous methods, and that's OK with me. Its just that this is simple
and easy to interpret, especially when there are outliers. Best
Harold Boxenbaum, Ph.D.
Arishel Inc. (Consulting)
Pharmacokinetic n Pharmaceutical n Medical n Biotechnology n
14621 Settlers Landing Way
North Potomac, MD 20878-4305
Phone: (301) 424-2806
Fax: (301) 424-8563
PharmPK Discussion List Archive Index page
Copyright 1995-2010 David W. A. Bourne (email@example.com)