0f512195d01

Upload: chevypickup

Post on 03-Apr-2018

217 views

Category:

Documents


0 download

TRANSCRIPT

  • 7/29/2019 0F512195d01

    1/14

    PLEASE SCROLL DOWN FOR ARTICLE

    This article was downloaded by:

    On: 19 January 2010

    Access details: Access Details: Free Access

    Publisher Taylor & Francis

    Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-

    41 Mortimer Street, London W1T 3JH, UK

    Journal of Biopharmaceutical StatisticsPublication details, including instructions for authors and subscription information:http://www.informaworld.com/smpp/title~content=t713597263

    Comparison of treatments in a combination therapy trialH. I. Patel aa Berlex Laboratories, Inc., Wayne, New Jersey

    To cite this Article Patel, H. I.(1991) 'Comparison of treatments in a combination therapy trial', Journal ofBiopharmaceutical Statistics, 1: 2, 171 183

    To link to this Article: DOI: 10.1080/10543409108835016URL: http://dx.doi.org/10.1080/10543409108835016

    Full terms and conditions of use: http://www.informaworld.com/terms-and-conditions-of-access.pdf

    This article may be used for research, teaching and private study purposes. Any substantial orsystematic reproduction, re-distribution, re-selling, loan or sub-licensing, systematic supply ordistribution in any form to anyone is expressly forbidden.

    The publisher does not give any warranty express or implied or make any representation that the contentswill be complete or accurate or up to date. The accuracy of any instructions, formulae and drug dosesshould be independently verified with primary sources. The publisher shall not be liable for any loss,actions, claims, proceedings, demand or costs or damages whatsoever or howsoever caused arising directlyor indirectly in connection with or arising out of the use of this material.

    http://www.informaworld.com/smpp/title~content=t713597263http://dx.doi.org/10.1080/10543409108835016http://www.informaworld.com/terms-and-conditions-of-access.pdfhttp://www.informaworld.com/terms-and-conditions-of-access.pdfhttp://dx.doi.org/10.1080/10543409108835016http://www.informaworld.com/smpp/title~content=t713597263
  • 7/29/2019 0F512195d01

    2/14

    K e y words. Restricted null hypothesis; Noncentral bivariate t ; Con-fidencc intcrva!; Op:i;r,a! allocatioc

    Abstract/>:+&,,,,;, ;h r 7;;;n i i . L --i i3'-,ir,d hi. 1 - - A ""-:"--- 1 \ E,..., ..,, ...... .:hi p.,,,,,, .,I ~ z j ~ dYICiji iCi : LQ: t e ~ t in casuperiority of a combination to each of its components is tile mi-. . . . . ,i'ormiy most powerfui nionoione iesi, i i is conservative in the neign-borhood o t origin. Considering a restricted nuii hypothesis, we ana--!yze a general h e a r rncde!. ! t: p viiiiie assockited ~ l i ? hhe prnposedtest and a confidence intervai for the shorTest distance between thecombination therapy and component therapies are computed. Anal-y s i s of a design with iwii or more iombinations is proposed. Anumerical exampie is given and a problem of optimai allocation ofthe sample sires anrd orher relevant design issues are discussed.

    Recen:!~, Laska an:! Pileisner ( 1) and Snapinn ( 2 ) ,among ~t h c r s , ave studieda problem of inference in a combination trial where a combination therapyis compared with its components, Suppose we restrict our attention to twocomponent therapies A and B and their combination C and to an efficacy

    Copyright 8 991 by Marcel Dekker, Inc.

  • 7/29/2019 0F512195d01

    3/14

    m.C L ' ; T ~ 6 -- - t. , . i Oti< Zt W ~ K < ~ t O i 73; { f , >

  • 7/29/2019 0F512195d01

    4/14

    Comparison of Treatnients in Combination Therapy Trial 173

    A Single Combination DesignIn this section we derive an asymptotic test for HiF given in Eq. (3 ) for the

    . * , \. ! ! - . . - / . > - A ?.-*.-I . . ": . - 3 - .g c ~ l G ~ ~ - tl i lca[ ~~l~~~~~d:yl - n r J aklu ; s { y j - r - i ~ I T nY LZ - ' < - , -15 V . LV l l l pULG.-.------*.- +I-nIIGpow" of th e test and a confi:?znce intervai for the shortest distance betweentine combination therapy and the components.

    Assuming normality for the distribution of y , we ,.btain the least squares(1s) estimator of ( 6 ; : 19,) and its dispersion matrix as ( 0 , . N1) and r = (y,,)d ~ ,espectively, where V = ; i f ) is a known matrix. The estimate of CT

    -1 .-

    is ci; = ( y ' y - ~ ' A ( A ' A ) ~ 'r j ) / v , based on v, degrees of freedom (d.i.).Let W, = i,/1/%nd p = ~ , , / \ ' ( v , , v ~ ~ ) ,here v,, = e2v,,. Under H z , the

  • 7/29/2019 0F512195d01

    5/14

    where J'yz,,2 j is tiie deiisiiy of Givariaie distri"uiiiion of the Stan-dardized variates Z, and Z2 with correlation p , ILJsinp quadrature, a SAS pro-gram is wriiter! to compute the bivarlate normal probability integral over acfinite or infinite rectangie. The program is given in Appeiadix 11. Note thatGupta ( 5 )has given extensive prvbabiiiiy tables for thi s distribution, bu t they:-sii.6-c 3,-,t7 '4 + p'7.cj az

  • 7/29/2019 0F512195d01

    6/14

    Numerical ExampleDiastolic blood pressure (mmHg) data were generated for a paraiiei-group- .dzsjgn ;wo incnorheraFizs anii their combicalioE.re!X rfpresenl h;iSE-line and :J the treatment response. Then for the three treatment groups, ran-dom samples of varying sizes were generated from the truncated bivariatenormal populations of ( X , Y) with means (+ = 102 , p i ) , = 1 , 2, 3 and thedispersion matrix (:;! ;;;j, where ,a,s the j t h element of the mean vector p i= (82, 87, 86). The last V i i i elemznts of p' represent the momtherapies 1and 2 and the first element represents the cornbinatioii. The x values weregenerated first from the marginal distribution of X over the range (90, 115)and then the y values were generated from the conditional distributions of Y

  • 7/29/2019 0F512195d01

    7/14

    Table I . Simulated Diastolic Blood Pressure ( ~ n m f i g ) at a arid Surnrrtary Statistics

    Mean 100.1 85.5 101.1S . D . 5.44 7.22 4.07

  • 7/29/2019 0F512195d01

    8/14

    Extension te a Desigr, with Two cr o re CambifiatiofisSuppose the early phase studies, expioratory in nature, s:aggest that a fewspecific dose combinations he frurther evaliuated i n comparison with theircomponents In a conrirnlatory study. A ruii factoriai design with K~K com-binations formed with R, doses of therapy i and R2 doses of therapy 2 hasa place in carly phases of the dmg deve!opmen: to understand the efficacyand safety profiles of the combination doses. Tnis, along with other existingknowledge for the class of drugs under investigation, should help choose asubset of combination doses for a more definitive study. Recently, Hung (7),

  • 7/29/2019 0F512195d01

    9/14

    !78 Paw!

    we wrife H,, as H,,, : C , @ (:, ' 8 = C2@ G , where @ stands for th e-.Kroneckei- product, 'l'tien for iarge vt., under H o ! , U = (C2fj)'

    /c 7L2A6 fir1\2 1 \L?L' / has 2 chi-squared distribr?tin Vvith2(( - !) d . , , whereC 2 8 is the Is estimator of C ' 2H , If i J is no t significant at a chosen level ofsignificance, one can proceed lo test the hypothesis H z . it is recommendedthat H,, be tested at the 0.2 or s o i n other level that is reasoriabfy higher tharithe conventional level uf 0.05. Such a high !eve! of significance for testingH,, can be justified because one cannot generally expect the same magnitudeso f 6 's in lo w and high dose combinations. When H,, is rejected, following

  • 7/29/2019 0F512195d01

    10/14

    Assuming that the response variable Y is normally distributed with knownVariai-,Ce ", suppose ii..e %ant to find an optima! a!!ocatim ~f the samplesizes n, , n2,an d ni with n, = n,) for designing a com bination therapy trial.Here x i and iz , 2re the srmr?!e sizes for th e cnmponent therapy groups an d I? : - m ax y, - d ) > 1 - tr (91I i

    ..A---* i l b i L ,I /-.n\.-"I-Ai i i i i ii, i i i i,- ,..-.,,-,.;c,,!.. .ii. . . -,,A - -r vL l --.lii--Uiiii 7, , i i , zed y , -.re t" s.;.-pls z p z n .fo r the ccmbination therapy and rhe component therapies I an d 2; respec-tively. With such an allocation, a lower ( 1 - a ) 100% confidence bound forthe shortest distance frox the csmhination theranv~ to the individual corn-ponents wiii be of width greater than or equal io d .The transformation V = -Y for th e response variable changes [Eq. (9)j

    to Eq. (2 .3) of Bechhofer and Tamhane (8) with P = 2 , y,, = ,x,,!n, A =d l v & / ~md t) = 2 . the^ fn!!owing the tables given by Rechhofer and T a m --hane (8), we can compute n! = n, and n,.. l o i i ius t ra te th is m ethod, iet ussuppose w e want to design a trial for the treatment of hypertension. Let cr =7 mmI-Cg, $ = 3 mii;Hg, and a = 0.05. Using 2 one-sided interval, va$mof -7" and A , corresponding to P = 2 and H = 2 ar e 9" = 6.399 and h =4 .651 . S ince ( ~ r r / d ) '= 11 7.8 , ii = 118. From this fi,. = 47 and f i , = ii, =36 .

  • 7/29/2019 0F512195d01

    11/14

    Acknowiedgments

    The author sincere!y thanks Professor A j ~ t . Tamhane, Nurthwesten Unl-versity, for his constructive suggestions and help in the preparation of thispaper. The author also wishes to thank an associate editor for his useful com-ments.

  • 7/29/2019 0F512195d01

    12/14

    Gamparisrr~ f Trrarmerrss in Combination Therap? Trial

    Conditioning on Yo = y , using the independence of Y, nd Y2 nd then un-cnnditinning on y , we get

    where @(. j is the cdf and &,) is the pGE of a N ( 0 , 1) random variable. Usingquadrature, one can evaluate the integral (for given x) first and then the ex-ternal integral.

  • 7/29/2019 0F512195d01

    13/14

    - . . * . . " . .' This t'njpi.am .iiiPiitcS blvi;ria;z r x i *r,> h.,p:!, i -- ?7tl : l i - : i j.liil:liil pi 5. ;e l bc q u a i iv 5 . 3.k .. 2 3;vfier, !oxjjer iirr.it e.: is .3: i ~ ~ S ~ A S i ~ S c 2 ~:L iaL : \1 .1~\3 3 :-,i 3iii: * 2: ::-* -+.;+ ::: ;- * .-'-:,g :**:r y -i' ".,.< :3 * y: :? >j

  • 7/29/2019 0F512195d01

    14/14