sequential PKPD modeling

From: "Jiang, Xuemin" xjiang1@medicine.bsd.uchicago.edu Subject: [NMusers] sequential PKPD modeling Date: Thu, 26 Oct 2006 09:53:07 -0500 Dear NONMEM users, I am developing PKPD modeling of a drug which can increase blood pressure (Drug A) simultaneously. A co-administered drug (Drug B) which can decrease blood pressure was found to increase the CL of Drug A. I am wondering whether the simultaneous PKPD modeling is able to confirm the finding or a sequential PKPD modeling need to be used in this case. Thanks Jim Jiang

From: "Jiang, Xuemin" xjiang1@medicine.bsd.uchicago.edu Subject: Re: [NMusers] sequential PKPD modeling Date: Thu, 26 Oct 2006 Dear NONMEM users, I would like to make my question clear. To my understanding, simultaneous analysis of PKPD data is a better analysis especially for the sparse PK data, because the PD data can support PK data analysis. I am wondering whether it is possible that the increasing on the CL of drug A (in the following email) could be a consequence of PD effect of drug B rather then a real PK effect in a simultaneous PKPD modeling. Jim Jiang

From: "Liping (CD) Zhang" liping.zhang3@bms.com Subject: Re: [NMusers] sequential PKPD modeling Date: Thu, 26 Oct 2006 15:54:50 -0400 Hi, Jim! "simultaneous analysis of PKPD data is a better analysis especially for the sparse PK data, because the PD data can support PK data analysis. " Simultaneous analysis of PKPD data is a better method because it naturally follows the maximal likelihood theory. PD's support for PK analysis is not for granted: a misspecified PD model could distorted PK model. For references, see Bennett and Wakefield, 2001, Biometrics, 57:803-12, and Zhang et al, 2003, JPP: 405-16. I assume you've had data from A alone, and data from A and B administrated, and when you model A+B PK/PD simultaneously, you get a large Cl for A than the one from A alone? (I could only think of this case, to support your statement "B was shown to increase the Cl of A"). If that is the case, doing the PK modeling on A using the data A+B, and then comparing the estimated Cl with estimated Cl from A alone, wouldn't that provide an answer? best regards, Liping

From: "Jiang, Xuemin" xjiang1@medicine.bsd.uchicago.edu Subject: Re: [NMusers] sequential PKPD modeling Date: Fri, 27 Oct 2006 09:41:12 -0500 Dear NONMEM users, Just would like to let you know that the drug B was a significant covariate on the CL of drug A while using simultaneous PKPD modeling but was not a significant covariate while using PK modeling only. On the other hand, I have very sparse PK data whether it is another reason the drug B is not significant. Many thank for the people replying the email Jim Jiang _______________________________________________________