Back to the Top
The following message was posted to: PharmPK
Hi everyone,
I realized the table I sent yesterday wasn't complete because it
didn't answer my original (and more pertinent question), how does the
proposed FDA method for scaled BE limits affect sample size
determination for a 2-way crossover BE study? I ran the numbers, and
the results are VERY interesting.
As you increase the IntraCV, the effect on sample size is positive
(more subjects required). However, as you widen BE limits, the effect
on sample size is negative, as less subjects are required to attain
the same amount of power. So which effect "wins out" with the proposed
FDA method, C2?
The answer is that the sample size initially lowers, then levels off
to about 33 completed subjects. The 4th column is the sample size
estimate for a 2-way BE study based on 80% power and a T/R ratio
between 95-105% with traditional BE limits (80.00-125.00%), and the
5th column is the sample size estimate assuming the same, using BE
limits scaled using the proposed Method C2.
[note to DB - please let me know if you'd like me to reformat this
table]
Sample Size
Estimate Sample Size Estimate
2-way
crossover 2-way crossover
IntraCV of IntraSD Scaled BE Limits 80% Power, T/
R=95% 80% Power, T/R=95%
Reference of Reference (Method C2) 80.00 - 125%
Limits Scaled BE Limits (C2)
30% 0.2936 80.00 - 125.00
39 39
32% 0.3122 78.88 - 126.78
44 38
34% 0.3307 77.77 - 128.58
49 37
36% 0.3491 76.70 - 130.38
54 36
38% 0.3673 75.64 - 132.20
60 36
40% 0.3853 74.62 - 134.02
66 36
42% 0.4031 73.61 - 135.84
72 35
44% 0.4207 72.63 - 137.68
78 35
46% 0.4381 71.68 - 139.51
84 35
48% 0.4553 70.75 - 141.35
91 34
50% 0.4724 69.84 - 143.19
98 34
52% 0.4892 68.95 - 145.04
105 34
54% 0.5058 68.08 - 146..88
112 34
56% 0.5223 67.24 - 148.73
119 34
58% 0.5385 66.41 - 150.57
126 33
60% 0.5545 65.61 - 152.42
134 33
62% 0.5703 64.83 - 154.26
142 33
64% 0.5859 64.06 - 156.10
149 33
66% 0.6013 63.32 - 157.94
157 33
68% 0.6165 62.59 - 159.77
165 33
70% 0.6315 61.88 - 161.60
173 33
72% 0.6463 61.19 - 163.42
181 33
74% 0.6608 60.52 - 165.24
189 33
76% 0.6752 59.86 - 167.06
198 33
78% 0.6894 59.22 - 168.87
206 33
80% 0..7033 58.59 - 170.67
214 33
The effect at 90% power with a T/R ratio between 95-105% is very
similar, starting at a sample size of n=52 (30% IntraCV), and
levelling off at n=41 (80% IntraCV).
If I were to play devil's advocate for Method C2, I would say that
there is a slight but noteworthy "reward" in power for having a higher
IntraCV if the ISV falls between 30 and 50% (very common for HVDs),
although the reward eventually wears off. However, the asymptotic
sample size (33 completed subjects) is a reasonable number for running
in the clinic.
Best,
--
David Dubins, Ph.D., B.Eng.
Global Bioequivalence Consulting
Assistant Professor, Leslie Dan Faculty of Pharmacy
University of Toronto
Back to the Top
David: this is interesting speculative work about a relevant issue:
please to avoid confusion, can you reformat the Table with discrete
columns and rows.
Angus McLean Ph.D.
8125 Langport Terrace,
Suite 100,
Gaithersburg,
MD 20877
E-mail angusmdmclean.aaa.aol.com
Back to the Top
The following message was posted to: PharmPK
Hi Dr. McLean,
I apologize for the formatting of the last table, for some reason the
width of the email was truncated when posted. Here is a "narrower"
table which should hopefully prevent wrap-around.
ISV SIGwr Scaled BE SS SS(C2)
30% 0.2936 80.00-125.00 39 39
32% 0.3122 78.88-126.78 44 38
34% 0.3307 77.77-128.58 49 37
36% 0.3491 76.70-130.38 54 36
38% 0.3673 75.64-132.20 60 36
40% 0.3853 74.62-134.02 66 36
42% 0.4031 73.61-135.84 72 35
44% 0.4207 72.63-137.68 78 35
46% 0.4381 71.68-139.51 84 35
48% 0.4553 70.75-141.35 91 34
50% 0.4724 69.84-143.19 98 34
52% 0.4892 68.95-145.04 105 34
54% 0.5058 68.08-146.88 112 34
56% 0.5223 67.24-148.73 119 34
58% 0.5385 66.41-150.57 126 33
60% 0.5545 65.61-152.42 134 33
62% 0.5703 64.83-154.26 142 33
64% 0.5859 64.06-156.10 149 33
66% 0.6013 63.32-157.94 157 33
68% 0.6165 62.59-159.77 165 33
70% 0.6315 61.88-161.60 173 33
72% 0.6463 61.19-163.42 181 33
74% 0.6608 60.52-165.24 189 33
76% 0.6752 59.86-167.06 198 33
78% 0.6894 59.22-168.87 206 33
80% 0.7033 58.59-170.67 214 33
Column Headers:
ISV: IntraCV of reference product (1st column)
SIGwr: Intrasubject standard deviation of reference product (2nd column)
Scaled BE: Bioequivalence limits scaled using the proposed FDA method
C2 (3rd column)
SS: Sample size using traditional 80-125% limits, given a T/R ratio
between 95-105% at 80% power (4th column)
SS(C2): Sample size using the scaled BE limits in the third column,
given a T/R ratio between 95-105% at 80% power (5th column)
Hope this helps,
David Dubins
[For future reference: Remember, when formatting tables etc. that
PharmPK messages are sent out (after conversion if necessary) as plain
text. Also some mail readers don't like long lines. David's table was
plain text but I gather the lines were a bit long for some - db]
Want to post a follow-up message on this topic?
If this link does not work with your browser send a follow-up message to PharmPK@boomer.org with "Proposed FDA Scaled BE Limits for HVDs: Effect on Sample Size Determination" as the subject | Support PharmPK by using the |
Copyright 1995-2011 David W. A. Bourne (david@boomer.org)