From: "Kyun-Seop Bae" ksbae@kma.org
Subject: [NMusers] Something wrong in US FDA guidance?
Date: Fri, 19 Nov 2004 17:06:50 +0900

Dear all,

I am asked to process an example data set in the US FDA "Guidance for
industry: Topical Dermatologic Corticosteroids in vivo Bioequivalence"
(http://www.fda.gov/cder/guidance/old098fn.pdf) using NONMEM.

In the gudiance, there is some data at page 28, and its results at page
29.
But, there exists quite difference between my result using NONMEM and
that of the guidance.

                           Guidance                             NONMEM
ED50                      1.89                                1.13 +/-
0.568
Emax                   -48.80                               -39.7 +/-
8.63


Of course, the confidence intervals embrace the values of guidance.
But, I think it's quite different, and different enough to lead
different action.

As the consequence of the analysis, they determined the doses(here, it
is duration, coded as TIME in my control file) to be tested are 1.0,
2.0, and 4.0 (for use in the pivotal study, adding half and double of
the ED50).
But, if we think NONMEM is better, the 'to be tested dose' should be
0.5, 1.0, and 2.0.

There is some points to consider, like in the guidance they did not use
NONMEM but 'P-Pharm'(I don't know much about this software), And,
curiously they seemed to use naive-pooling method, even though they said
that P-Pharm is population PK-PD data modeling software. 

I tried also various error, eta, and covariance models, and even
naive-pooling method useing NONMEM.
But all the results are very similar to the above NONMEM result.   

For your convenience, I have included the NONMEM control code, data, and
some graphics.

Please, take a look and if you think my result is correct,
give me some explanations to make others(non-NONMEM users) understand.

Sincerely,

Kyun-Seop Bae MD PhD

Department of Clinical Pharmacology &
Clinical Research Center
Asan Medical Center

388-1 Pungnap-dong
Songpa-gu, Seoul 138-736, Korea
Tel: +82-2-3010-5901
Fax: +82-2-3010-5509
Email: ksbae@amc.seoul.kr
==========================================

[Control File]
$PROBLEM Skin PD
$INPUT ID TIME DV MDV
$DATA Steroid.CSV    IGNORE=#
$PRED
  EMAX = THETA(1)*EXP(ETA(1))
  EC50 = THETA(2)*EXP(ETA(2))
 
  IPRE  = 0 - EMAX * TIME / (EC50 + TIME)
  Y     = IPRE +  EPS(1)
 
$THETA
  (0, 55)
  (0, 2.5)
 
$OMEGA
  0.2
  0.2
 
$SIGMA 100
$ESTIMATION MAXEVAL=9999 PRINT=5 METHOD=ZERO POSTHOC
$COVARIANCE


[Data File]
#Subject,Dose Duration(hr),AUEC,MDV
1,0.25,-1.23,0
1,0.5,-7.39,0
1,0.75,-1.48,0
1,1,-3.8,0
1,1.5,-0.23,0
1,2,5.77,0
1,4,-4.74,0
1,6,-1.53,0
2,0.25,-0.02,0
2,0.5,-5.13,0
2,0.75,-8.92,0
2,1,-24.56,0
2,1.5,-19.21,0
2,2,-1.8,0
2,4,-43.07,0
2,6,-41.56,0
3,0.25,-13.87,0
3,0.5,-15.03,0
3,0.75,-18.39,0
3,1,-16.25,0
3,1.5,-15.44,0
3,2,-23.74,0
3,4,-24.8,0
3,6,-21.79,0
4,0.25,-27.27,0
4,0.5,-3.71,0
4,0.75,-43.82,0
4,1,-44.39,0
4,1.5,-77.04,0
4,2,-66.8,0
4,4,-62.96,0
4,6,-71.6,0
5,0.25,-10.65,0
5,0.5,7.72,0
5,0.75,-23.42,0
5,1,-20.37,0
5,1.5,-19.95,0
5,2,-32,0
5,4,-32.81,0
5,6,-61.51,0
6,0.25,-10.41,0
6,0.5,-5.94,0
6,0.75,-2.29,0
6,1,-8.92,0
6,1.5,-20.64,0
6,2,-19.52,0
6,4,-8.52,0
6,6,-19.01,0
7,0.25,4.2,0
7,0.5,-12.31,0
7,0.75,1.34,0
7,1,-18.84,0
7,1.5,-42.7,0
7,2,-37.29,0
7,4,-45.46,0
7,6,-37.24,0
8,0.25,-11.95,0
8,0.5,7.45,0
8,0.75,5.95,0
8,1,8.78,0
8,1.5,1.26,0
8,2,-48.83,0
8,4,-71.77,0
8,6,-8.14,0
9,0.25,-12.36,0
9,0.5,12.95,0
9,0.75,1.88,0
9,1,-43.35,0
9,1.5,-20.97,0
9,2,-39.79,0
9,4,-57.55,0
9,6,-34.18,0
10,0.25,1.15,0
10,0.5,-39.45,0
10,0.75,-40.68,0
10,1,-16.19,0
10,1.5,6.87,0
10,2,10.75,0
10,4,-37.64,0
10,6,-35.01,0
11,0.25,-30.03,0
11,0.5,-39.56,0
11,0.75,-61.06,0
11,1,-43.58,0
11,1.5,-40.73,0
11,2,-62.01,0
11,4,-27.82,0
11,6,-33.6,0
12,0.25,-7.25,0
12,0.5,14.73,0
12,0.75,-21.09,0
12,1,10.81,0
12,1.5,0.51,0
12,2,-10.51,0
12,4,-14.89,0
12,6,16.14,0
 

  _____  


================================

Kyun-Seop Bae MD PhD
Clinical Research Center,  Asan Medical Center
Pungnap-dong Songpa-gu  Seoul, 138-736, Republic of Korea
Tel: +82-2-3010-5901  Fax: +82-2-3010-5509  Email: ksbae@kma.org

_______________________________________________________

From: "Bachman, William (MYD)" bachmanw@iconus.com
Subject: RE: [NMusers] Something wrong in US FDA guidance?
Date:  Fri, November 19, 2004 8:03 am 

Kyun-Seop,

I am not surprised that two different method gives two different answers.
For an explanation of the differences between NONMEM and P-Pharm you may
want to read:

"J Biopharm Stat. 1995 Jul;5(2):141-58.   A two-step iterative algorithm for
estimation in nonlinear mixed-effect models with an evaluation in population
pharmacokinetics.

Mentre F, Gomeni R.
INSERM U194, Service de Biostatistique et Informatique Medicale, CHU
Pitie-Salpetriere, Paris, France.

This article proposes an EM-like algorithm for estimating, by maximum
likelihood, the population parameters of a nonlinear mixed-effect model
given sparse individual data. The first step involves Bayesian estimation of
the individual parameters. During the second step, population parameters are
estimated using a linearization about those Bayesian estimates. This
algorithm (implemented in P-PHARM) is evaluated on simulated data, mimicking
pharmacokinetic analyses and compared to the First-Order method and the
First-Order Conditional Estimates method (both implemented in NONMEM). The
accuracy of the results, within few iterations, shows the estimation
capabilities of the proposed approach."

Perhaps someone at the FDA or elsewhere has repeated the analysis with
NONMEM and will share their results?

nmconsult@globomaxnm.com
GloboMax®
The Strategic Pharmaceutical Development Division of ICON plc
7250 Parkway Drive, Suite 430
Hanover, MD 21076 
Voice: (410) 782-2205
FAX: (410) 712-0737
_______________________________________________________

From: "Simon Davis"  
Subject: RE: [NMusers] Something wrong in US FDA guidance?
Date: Fri, November 19, 2004 9:02 am 

Kyun-Seop,
        Bill is correct the EM algorithm is the one used by P-Pharm; in
2000 this algorithm was incorporated into InnaPhase's standard PK
software Kinetica.  This enables users to design and perform NCA,
compartmental and population analysis through a common interface.

        If you would like to access a free demonstration download of
Kinetica, please contact me off the list and I will be happy to explain
more.  This offer is open to anyone else who is interested in comparing
the two applications further.

http://www.innaphase.com/support_downloads_kinetica.html

   Best regards,
                            Simon.
___________________________________
Simon Davis (skype = simon_davis)
Senior Technical Support Scientist, Europe
Thermo Electron Corporation:  www.innaphase.com

Cell phone : +44 7980 832 666
Facsimile  : +1 801 991 7145
Office phone : +44 161 942 3000

13 Sep 2004 Thermo Electron Corporation, the world leader in analytical
instruments, scientific equipment, software and services, acquired
InnaPhase, bolstering the company's software offering to the life
sciences industry and creating the world leading LIMS vendor.
http://phx.corporate-ir.net/phoenix.zhtml?c=89145&p=irol-newsArticle&t=Regular&id=612672
_______________________________________________________