Download Title: Dosing Strategies of Drugs with Narrow Therapeutic Windows

Title: Dosing Strategies of Drugs with Narrow Therapeutic Windows: A simple Approach to
Determine Trade-off Between Efficacy and Toxicity
Authors: Mohamad-Samer Mouksassi* (1,2), Chunlin Chen (1,2), Jean-François Marier (1)
Institutions: (1) Pharsight Corporation, Reporting and Analysis Services, Montreal, Canada
(2) Université de Montréal, Faculté de pharmacie, Montreal, Canada
Objectives: Dosing strategies of drugs with narrow therapeutic windows are difficult to develop and apply in
clinical practice. Population pharmacokinetics offers the possibility of explaining the between subject variability
(BSV) with patient-specific covariates. The objective was to develop a quick method to determine trade-offs
between efficacy and toxicity in order to design better dosing strategies given a BSV of a drug with narrow
therapeutic window.
Methods: The methodology will be illustrated using a Drug X, with a pre-defined AUC therapeutic window of
220 and 360 mg.h/L. Assuming that a population pharmacokinetics analyses showed that BSA is a significant
covariate on clearance (Power model, CL=TVCL.(BSA/1.1)1.2), and that the unexplained BSV on AUC was 38%,
an original dosing regimen of 0.8 and 1.0 mg/m2 was developed for patients with BSA < 1.2 m2 and > 1.2 m2,
respectively. This dosing regimen resulted in a median AUC of 200 mg.h/L and a percentage of patients in the
sub-therapeutic, therapeutic and toxic range of 53%, 41% and 6%, respectively. The first method (Method I) to
estimate an optimal dosing strategy was previously described by Jonsson et al.1 Optimal cutoffs and doses to use
were estimated with 2, 3, and 5 BSA cutoffs. The loss function was chosen to be quadratic, symmetrical and
centered on the middle of the therapeutic window (i.e: 290 mg.h/L). Each estimated dosing strategy was then
tested by simulation (n=200), incorporating parameter uncertainty, and the percentage of patients within subtherapeutic, therapeutic and toxic ranges were computed. The second method (Method II) used the area under the
log-normal distribution of AUC. The dosing strategy was assumed to change the location parameter (translation
on the x-axis) but not the variance. Percentage of the patient population within sub-therapeutic, therapeutic and
toxic ranges were computed and plotted versus the targeted AUC. The optimal AUC was deduced graphically as a
trade-off between the population of patients in therapeutic and toxic ranges.
Results: For Method I, optimal cutoffs for 2, 3, and 5 dose levels the percentages of patients within the subtherapeutic, therapeutic and toxic ranges are presented in Figure 1. As the number of cutoffs increased, the
distribution of AUC was more centered on the target (i.e., the middle of the therapeutic range). However, because
of the large variability, the percentage of patients in the toxic range did not necessarily decrease.
RESULTS FOR 200 SIMULATIONS FIVE DOSES REGIMEN
RESULTS FOR 200 SIMULATIONS ORIGINAL DOSE REGIMEN
46%
48%
6%
56%
19%
Density
Density
25%
0
100
200
100
UNDERDOSED
300
AUC THERAPEUTIC
400
500
TOXIC
6
19
80
24
60
48
56
20
40
55
46
25
0
21
2
3
Number of Doses
5
0
100
200
300
400
500
AUC
Figure 1: Upper left, densities of AUC with the
original dosing regimen. Upper right, densities of AUC
with 5 dosing levels. Lower Left, barplot summarizing
the results of the different regimens tested. Green areas
display the therapeutic window or percentage of
therapeutic patients. Blue and red areas display subtherapeutic and toxic areas or patient populations
within sub-therapeutic or toxic ranges, respectively.
Figure 2: The green line (therapeutic
patients), the blue line (sub-therapeutic
patients) and the red line (toxic patients)
versus a range of possible target AUC’s. A
red horizontal line was drawn at 10%. The
target should be at most 220 mg.h/L in order
to keep the percentage of patients with
toxicity below 10%. The current targeted
AUC would result in 40% of the population
in the therapeutic zone (green horizontal line)
0%
Percentage of patients
10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
Method II calculated the percentages of patients within the sub-therapeutic, therapeutic and toxic range as a
function of the target AUC (Figure 2). This approach helps to visualize the percentage of patients in the
therapeutic and toxic range for any targeted AUC. Furthermore, this helps to optimize the targeted AUC
according to pre-defined acceptable percentage of patients within the toxic range. Figure 2 shows that a target
AUC of 290 mg.h/L would result in the highest percentage of patients within the therapeutics range while toxicity
would increase to more than 20%. This is consistent with the results of Method 1. An alternative to Method II
would be to conduct Method I with more complex loss function with a penalty if a patient falls within the toxic
range.
100 120 140 160 180 200 220 240 260 280 300 320 340 360 380 400
Target AUC
Conclusions: Method I and II are complementary. Before conducting Method I, we suggest to determine
graphically the best targeted AUC in order to determine the trade-off between patients in therapeutic and toxic
ranges. Once an optimal target AUC is determined, Method I can be used to estimate the optimal cutoffs and
doses that will center the distribution of AUC over the optimal target.
References:
[1] Jonsson S, Karlsson MO. A rational approach for selection of optimal covariate-based dosing strategies.
Clin Pharmacol Ther. 2003 Jan;73(1):7-19.