Clinical Trial Simulation using NONMEM

Slide 1 Clinical Trial Simulation Using NONMEM Nick Holford Dept Pharmacology & Clinical Pharmacology University of Auckland, New Zealand ©NHG Holford, 2012, all rights reserved. Slide 2 NONMEM Example $PROB THEOPHYLLINE PHARMACODYNAMICS $PRED $DATA theopd.dat IGNORE # S0=POP_S0*EXP(PPV_S0) $INPUT ID TIME THEO AGE WT GEND RACE DIAG DV EMAX=POP_EMAX*EXP(PPV_EMAX) $SIM (20000625 NORMAL NEW) SUBPROBLEMS=100 $ESTIM PRINT=0 ; suppress output $THETA (0,150.,) ; POP_S0 $THETA (0,200.,) ; POP_EMAX C50=POP_C50*EXP(PPV_C50) EFFECT=S0 + EMAX*THEO/(THEO+C50) Y = EFFECT + RUV_SD IF (ICALL.EQ.4) THEN DOWHILE (Y.LT.0) CALL SIMEPS(EPS) $THETA (.001,10,) ; POP_C50 Y=EFFECT + RUV_SD $OMEGA BLOCK(3) 0.25 ENDDO ; PPV_S0 0.01 0.25 ; PPV_EMAX 0.01 0.01 0.25 ENDIF ; PPV_C50 $SIGMA 100 ; RUV_SD ©NHG Holford, 2012, all rights reserved. Slide 3 Simulating Continuous Covariates IF (ICALL.EQ.4) THEN ; Simulation ; Simulate Weight Distribution (for male) WTMALE=WTSTD*EXP(PPV_WT) LO=40 ; kg HI=200 ; kg DOWHILE (WTMALE.LT.LO .OR. WTMALE.GT.HI) CALL SIMETA(ETA) WTMALE = WTSTD*EXP(PPV_WT) ENDDO etc ©NHG Holford, 2012, all rights reserved. Slide 4 Simulating Categorical Covariates $SIM (20000625 NORMAL NEW) (20010112 UNIFORM) SUBPROBLEMS=100 $PK ; Simulate Sex Distribution FEMALE=0.5 ; proportion of females CALL RANDOM(2,R) ; 2nd random number generator IF (R.GT.FEMALE) THEN SEX=1 ;male WT=WTMALE ; previously simulated male WT ELSE SEX=0 ;female WT=0.85*WTMALE ; weight for female ENDIF etc ©NHG Holford, 2012, all rights reserved. Slide 5 Truncating Parameters IF (ICALL.EQ.4) THEN ; simulation within 99.9% of full Normal TRUNC=3.27 ; Z 2tailed alpha=0.01 i.e. include 99.9% GRPS0=POP_S0 ; include covariate effects here if needed S0=GRPE0*EXP(PPV_E0) LNMU=LOG(GRP_E0) DLTA=TRUNC*0.717 ; Must be TRUNC*SQRT(PPV_E0)! LO=EXP(LNMU-DLTA) HI=EXP(LNMU+DLTA) DOWHILE (S0.LT.LO .OR. S0.GT.HI) CALL SIMETA(ETA) S0=GRPS0*EXP(PPV_S0) ENDDO etc ©NHG Holford, 2012, all rights reserved. Slide 6 Hands-On • Original data from RCCT of theophylline • Use original data to obtain model parameters • Simulate a Randomized Concentration Controlled Trial using NONMEM • Evaluate model analysis scenarios – Placebo vs Target (10 or 20 mg/L) – Placebo vs Actual Concentration Holford N, Black P, Couch R, Kennedy J, Briant R. Theophylline target concentration in severe airways obstruction - 10 or 20 mg/L? A randomised concentration-controlled trial. Clin Pharmacokinet 1993; 25: 495-505. ©NHG Holford, 2012, all rights reserved. Slide 7 Fit Original Data ©NHG Holford, 2012, all rights reserved. Slide 8 Simulate Data ©NHG Holford, 2012, all rights reserved. Slide 9 Examine Simulation In Excel ©NHG Holford, 2012, all rights reserved. Slide 10 No Effect Estimation $PROB theophylline concentration controlled trial $DATA ..\theopd_sim_trunc.reg\theopd_sim_trunc.fit IGNORE @ $INPUT ID TRT TIME CONC DV MDV $ESTIM METHOD=COND NSIG=3 SIGL=9 $THETA (0,150.,) ; POP_S0 L/min (0,10,) ; RUV_SD L/min $OMEGA BLOCK(1) 0.25 ; PPV_S0 $SIGMA 1 FIX ; EPS1 $PRED S0=POP_S0*EXP(PPV_S0) Y = S0 + RUV_SD*EPS1 ©NHG Holford, 2012, all rights reserved. Slide 11 Treatment Target Estimation $PROB theophylline concentration controlled trial $DATA ..\theopd_sim_trunc.reg\theopd_sim_trunc.fit IGNORE @ $INPUT ID TRT TIME CONC DV MDV $ESTIM METHOD=COND NSIG=3 SIGL=9 $THETA (0,150.,) ; POP_S0 L/min 20 ; POP_EFF10 L/min 20 ; POP_EFF20 L/min (0,10,) ; RUV_SD $OMEGA BLOCK(2) 0.25 ; PPV_S0 0.01 0.25 ; PPV_EFFECT $SIGMA 1 FIX ; EPS1 $PRED S0=POP_S0*EXP(PPV_S0) IF (TRT.EQ.10) THEN EFFECT=POP_EFF10*(1+PPV_EFFECT) ELSE EFFECT=POP_EFF20*(1+PPV_EFFECT) ENDIF Y = S0 + EFFECT + RUV_SD*EPS1 ©NHG Holford, 2012, all rights reserved. Slide 12 Actual Concentration Estimation $PROB theophylline concentration controlled trial $DATA ..\theopd_sim_trunc.reg\theopd_sim_trunc.fit IGNORE @ $INPUT ID TRT TIME CONC DV MDV $ESTIM METHOD=COND NSIG=3 SIGL=9 $THETA (0,150.,) ; POP_S0 L/min 200. ; POP_EMAX L/min (.1,10,); POP_C50 mg/L (0,10,) ; RUV_SD L/min $PRED $OMEGA BLOCK(3) 0.25 ; PPV_S0 0.01 0.25 ; PPV_EMAX 0.01 0.01 0.25 ; PPV_C50 S0=POP_S0*EXP(PPV_S0) EMAX=POP_EMAX*(1+PPV_EMAX) C50=POP_C50*EXP(PPV_C50) $SIGMA 1 FIX EFFECT= S0 + EMAX*CONC/(CONC+C50) Y = EFFECT + RUV_SD*EPS1 ©NHG Holford, 2012, all rights reserved. ; EPS1 Slide 13 Simulate and Estimate rem rem rem rem rem To create and delete simulated data: set To create and keep simulated data: set To skip creation and keep sim data: set Non-default simulated data dir: set create and keep simulated data, estimate ctsthisgotdata= ctsthisgotdata=n ctsthisgotdata=y ctsdata=non_default_dir with placebo model set ctsthisgotdata=n call nmgosim theopd_sim_trunc trial_placebo_est 1 10 rem use simulated data, estimate with treatment model set ctsthisgotdata=y call nmgosim theopd_sim_trunc trial_trt_est 1 10 rem use simulated data, estimate with concentration model set ctsthisgotdata=y call nmgosim theopd_sim_trunc trial_conc_est 1 10 ©NHG Holford, 2012, all rights reserved. Slide 14 Power Obj Obj DeltaConc df 4172.766 4165.529 7.237 4095.863 4086.683 9.18 4060.57 4051.011 9.559 4083.553 4077.46 6.093 4073.254 4060.229 13.025 4136.384 4134.839 1.545 4088.738 4057.031 31.707 4152.024 4137.328 14.696 4137.879 4124.76 13.119 4096.63 4077.816 18.814 4 4 4 4 4 4 4 4 4 4 P Significant 0.123881474 0 0.056754478 0 0.048548944 1 0.192309795 0 0.011154283 1 0.8186415 0 2.19585E-06 1 0.0053751 1 0.010708741 1 0.000854904 1 Obj Obj DeltaConc df 4172.766 4157.233 15.533 4095.863 4079.548 16.315 4060.57 4042.434 18.136 4083.553 4056.732 26.821 4073.254 4054.526 18.728 4136.384 4120.128 16.256 4088.738 4067.576 21.162 4152.024 4131.046 20.978 4137.879 4119.766 18.113 4096.63 4065.427 31.203 7 7 7 7 7 7 7 7 7 7 P 0.02974275 0.022389236 0.011371235 0.000358927 0.009083485 0.022877696 0.003537637 0.003802848 0.011470439 5.70359E-05 Significant 1 1 1 1 1 1 1 1 1 1 Treatment Target alpha 0.05 power 60% Actual Concentration alpha 0.05 power 100% ©NHG Holford, 2012, all rights reserved. Slide 15 Simulation Tools • NONMEM – Easy to simulate from estimation model • Wings for NONMEM – Support for NONMEM estimation to test scenarios • Excel – Graphical check of simulation – Power Calculation ©NHG Holford, 2012, all rights reserved.