From: Johan.Rosenborg@astrazeneca.com Subject: [NMusers] Concentration dependent volume of distribution Date: Monday, 18 March 2002 17:47 All, I would like to model concentration or possibly amount dependent volume of distribution, using ADVAN8. Does anyone have a suggestion on how to deal with this problem? I have tried to introduce VD, VMAX and A50 as shown below, but it did not work. / Johan $PROBLEM $INPUT ID AMT SS II RATE TIME EVID CMT DV FLAG STUD $DATA dat.prn IGNORE=# $SUBROUTINES ADVAN8 TOL=4 $MODEL COMP=(CENTRAL) COMP=(PERIPHL) COMP=(PERIPHC) $PK K10 =THETA(1)*(1+ETA(1)) K12 =THETA(2) K21 =THETA(3) K13 =THETA(4)*(1+ETA(2)) K31 =THETA(5) ;------------ Volume------------------------- VD =THETA(6)*(1+ETA(3)) VMAX=THETA(9)*(1+ETA(5)) A50 =THETA(10)*(1+ETA(6)) ;-------------------------------------------- F0 =THETA(7)*(1+ETA(4)) $DES S1 =VD+VMAX*A(1)/(A50+A(1)) C1 =A(1)/S1 DADT(1)=K41*A(4)-(K10+K12+K13)*C1*S1+K21*A(2)+K31*A(3) DADT(2)=K12*C1*S1-K21*A(2) DADT(3)=K13*C1*S1-K31*A(3) $ERROR IF (EVID.EQ.0.AND.F.GT.0) THEN IPRED=LOG(F) ELSE IPRED=LOG(F+0.001) ENDIF IRES=DV-IPRED IWRES=IRES X6=0 X7=0 IF(CMT.EQ.1)X6=1 IF(CMT.EQ.(-5))X7=1 Y=IPRED+X6*ERR(1)+X7*ERR(2) ******* From: Piotrovskij, Vladimir [PRDBE] [mailto:VPIOTROV@PRDBE.jnj.com] Subject: RE: [NMusers] Concentration dependent volume of distribution Date: Tuesday, March 19, 2002 5:04 AM Johan, I don't think making V explicitly dependent on conc of amount is a right way to follow. If V is concentration-dependent it indicates nonlinear distribution caused by, eg, saturable protein binding. You'd better explore nonlinear binding in the central or peripheral compartment, or, alternatively, a saturable transport to (one of the) peripheral compartments, etc. Best regards, Vladimir ******* From: "Wang, Bing"Subject: FW: [NMusers] Concentration dependent volume of distribution Date: Tue, 19 Mar 2002 09:45:44 -0800 John, This is a loop - the calculation of Conc depends on Vc while in your model Vc is Conc-dependent. It will confuse NONMEM... Most likely the observed conc-dependent Vc is caused by the nonlinear protein binding at (extremely?) low dose levels. A saturable binding compartment (e.g., COMP 2) may be introduced: K12=KON*(RMAX-A(2)) ; RMAX is the maximum number of binding sites. K21=KOFF .... then use K12 and K21 in your DES. It is possible that you have to fix KOFF=1 to avoid overparameterization... Regards. Bing Wang ******* From: "Bachman, William" Subject: RE: [NMusers] Concentration dependent volume of distribution Date: Tue, 19 Mar 2002 11:08:23 -0500 Johan, Aside from Vladimir's biological objections (which have merits), from a strictly modeling viewpoint, you seem to have an undefined fourth compartment (as evidenced by DADT(1)=K41*A(4) ... ) unless I'm missing something here. What is this term supposed to represent? Bill *******