separating the dynamic components of placebo effect, disease … · 2018-06-07 · literature model...
TRANSCRIPT
Literature Model for FEV1 in COPD Trials Separating the Dynamic Components of Placebo Effect,
Disease Progression and Interacting Drug Effects
GlobalPharmacometrics
Jakob Ribbing, Christine Falcoz,Itzela Correa and Steven W Martin
21st PAGE meeting, Venice (Italy)6th June 2012
Acknowledgements
• Lutz Harnisch
• Margherita Bennetts
• Tracy Higgins
• Thomas Kerbusch
• John C Lukas
2
The FEV1 biomarker in COPD
• Chronic obstructive pulmonary disease (COPD)– Third leading cause of death in US and
projected to increase world-wide due toprojected to increase world wide due to smoking
• Forced expiratory volume in one second (FEV1)– Important endpoint for diagnosis and the
primary biomarker for dose selection in ph2b– Ph3 need to show reduced exacerbations 3
Example of how literature analysis aids internal development at PfizerInternal Data Literature Data
FEV1
Model prediction of Drugs X/Y Dose-
Response in FEV1
Model prediction of FEV1-treatment effects
across published compounds
FEV1 comparison to competitor drugs (MDI)
4
Exac
erba
tion
rate
(ER
) Model prediction of FEV1-ER relation across
published compounds
Predicted efficacy in ER across published
compounds
Model prediction of Drugs X/Y Dose-Response in ER
ER comparison to competitor drugs (MDI)
Drug Class of interest in COPD
• Long-acting bronchodilators (inhaled)– Long-acting β2 agonists (LABA)
L ti ti h li i (LAAC)– Long-acting anticholinergics (LAAC)
• Anti-inflammatory therapy– Inhaled corticosteroids (ICS)– PDE4-inhibitors (PDE4i)
5
Literature FEV1 in COPD Change from baseline trough FEV1 (up to wk 26)
by treatment classICS ICS LABA ICS LABA LAAC
0.0
0.1
0.2
LAAC
FB (L
)
Absolute FEV1 is
analysed, but displayed as
6
0.0
0.1
0.2
0 5 10 15 20 25
LABA LABA LAAC
0 5 10 15 20 25
PBO PDE4i
Time (weeks)
Trou
gh F
EV
1 C
F displayed as change from baseline, due
to dominating variability in
baseline!
Linear disease progressionDisease modifying effects?
-0.2
-0.1
0.0
Pauwels (1999) Ref=637 Trial ID=43 N=1277 Post-BD
1 1 11 1 1
11
1 1 11
1
-0.1
0.0
0.1
Tashkin (2008) NCT00144339 Ref=661 Trial ID=75 N=5993 Pre-BD
1
1 1 11
11 1 1
1
UPLIFT study
7
0 50 100 150
-0.3
PBOBudesonide 800 UG/DAY 1
Time (weeks)
CFB
FEV
1 (L
)
0 50 100 150 200-0
.2-
PBOTiotropium 18 UG/DAY 1
Time (weeks)
CFB
FEV
1 (L
)
0 50 100 150
-0.2
-0.1
0.0
0.1
Celli (2008) NCT00268216, SCO30003 Ref=663 Trial ID=81 N=6112 Post-BD
PBO
1
1 11
1 11
Fluticasone/Salmeterol 1000/100 UG/DAY 1
2
22 2
22
2
Fluticasone 1000 UG/DAY 2
3
33 3
33
3
Salmeterol 100 UG/DAY 3
Time (weeks)
CFB
FEV
1 (L
)
0 50 100 150 200
-0.3
-0.2
-0.1
0.0
UNK (2000) Ref=741 Trial ID=60 N=1116 Pre-BD
PBO
1 1
11
11
11
1
Triamcinolone 1200 UG/DAY 1
Time (weeks)
CFB
FEV
1 (L
)
48 / LIKE COPD 110621 l / f 1 19 / 04J 2012
y
TORCH study
Week
CFB
in F
EV
1(L)
Illustration of ISV in placebo effect at typical disease progression
1.22
Short study duration Long study duration
Pre-SABA/SAAC
Placebo95% Conf. Int.
Legend
FEV1active = Baseline + Effplacebo + Progressiondisease
Graph illustrates gradual placebo
8Week
FEV
1 (L
)
1.18
1.19
1.20
1.21
0 1 2 3 4 5 6
Year
1.05
1.10
1.15
1.20
0 1 2 3 4
gradual placebo response, but
mixture component allowed immediate placebo response
as well
Placebo, Salmeterol (LABA) and Fluticasone (ICS)
1.40
Short study duration1.4
Long study duration
Pre-SABA/SAAC
PlaceboSalmeterol 50 mcg BIDFluticasone 250 mcg BID
Legend
FEV1active = Baseline + Effplacebo + Progressiondisease + Effdrug
9Week
FEV
1 (L
)
1.20
1.25
1.30
1.35
0 1 2 3 4 5 6
Year
1.1
1.2
1.3
0 1 2 3 4
Salmeterol interaction with FEV1 post-SABA/SAAC
1.40
Short study duration1.4
Long study duration
Pre-SABA/SAAC
PlaceboSalmeterol 50 mcg BIDFluticasone 250 mcg BID
Post-SABA/SAACLegend
FEV1active = Baseline + Effplacebo + Progressiondisease + Effdrug
10Week
FEV
1 (L
)
1.20
1.25
1.30
1.35
0 1 2 3 4 5 6
Year
1.1
1.2
1.3
0 1 2 3 4
Covariates
• Pre-specified– baseline parameter, based on inclusion criteria:
• Min/max disease severity:mild/moderate/severe/v. severe• Restricted medical history limits more severe patients• min/max #exacerbations in previous year (no, any#, ≥1, ≥2)
– Disease prog. proportional to (ipred) baselineuntreated
• Selected at p<0.001:– Baseline decline with Age– Baseline < 1.2 L: Linear decline in anti-inflammatory
efficacy11
η-diagnostics: Base modelBaseline affects drug response
typi
cal e
ffica
cy
1.2
1.4ETA on AI efficacy
1.0 1.5 2.0ETA on BD efficacy
12
Untreated study baseline (L)
Rat
io: i
pred
effi
cacy
ove
r
0.4
0.6
0.8
1.0
1.0 1.5 2.0
η-diagnostics: Base modelBaseline affects drug response
typi
cal e
ffica
cy
1.2
1.4 fluticasone trialsbudesonide trials
roflumilast trials599
648
637165663 668684
714577308
261
ETA on AI efficacy
1.0 1.5 2.0
salmeterol trialsformoterol trialstiotropium trials
indacaterol trials693
353573490
ETA on BD efficacy
13
Untreated study baseline (L)
Rat
io: i
pred
effi
cacy
ove
r
0.4
0.6
0.8
1.0
1.0 1.5 2.0
785649
649
648573750
679
657
647
609
695 664
668
779
786708706
666164 251
270
288328 509 609621695 664
663 668779684708706
714577308709
353
605
639
610261
573 653659
660673690 776778
807
679657
678
664647 656 1536536 765
251 607658767686
683
288455488492493
494
621
653659
660 662673690695
762676
661664
668779 687647 656 667694807798790
ETA diagnostics: Final modelty
pica
l effi
cacy
1 05
1.10
1.15 fluticasone trialsbudesonide trials
roflumilast trials649 648
663261
ETA on AI efficacy
1.0 1.5 2.0
salmeterol trialsformoterol trialstiotropium trials
indacaterol trials
693
684714
353639573
653807536 765
490
494653676 807
ETA on BD efficacy
14
Untreated study baseline (L)
Rat
io: i
pred
effi
cacy
ove
r
0.85
0.90
0.95
1.00
1.05
1.0 1.5 2.0
599
785649
648
573 637750
679
657
647
165
609695
664
668779
684
786708706
714666577308
164
261
251
270
288328509
609
621695
664
663 668779
708706
714
577
308
709605
610261659
660
673690776
778
807
679657
678
664
647656
1536
765
251 607
658767
686
683
288455488492
493621659
660
662
673690
695762
676
661
664
668779 687647
656 667
694807
798
790
Efficacy in moderate COPD
formoterol9 mcg BID
indacaterol75 mcg QD
SX 50mcg BID andSpiriva 18 mcg QD
SXFP 50/250 mcg BID
86 ml
149 ml
162 ml
154 ml
Pre-SABA/SAAC
LABA + ICS same efficacy as the sum
of the two mono components
15Placebo-corrected FEV1 (L)
roflumilast500 mcg QD
budesonide400/320 mcg BID
Fluticasone250 mcg BID
Spiriva18 mcg QD
Respimat5 mcg QD
salmeterol50 mcg BID
0.05 0.10 0.15
36 ml
61 ml
83 ml
92 ml
128 ml
116 ml
Efficacy in moderate COPD
formoterol9 mcg BID
indacaterol75 mcg QD
SX 50mcg BID andSpiriva 18 mcg QD
SXFP 50/250 mcg BID
86 ml
149 ml
162 ml
154 ml
Pre-SABA/SAAC
LABA+LAACinteraction estimated as a relative reduction
in LABA effect
16Placebo-corrected FEV1 (L)
roflumilast500 mcg QD
budesonide400/320 mcg BID
Fluticasone250 mcg BID
Spiriva18 mcg QD
Respimat5 mcg QD
salmeterol50 mcg BID
0.05 0.10 0.15
36 ml
61 ml
83 ml
92 ml
128 ml
116 ml
Efficacy in moderate COPD95% CI excluding treatment heterogeneity
Indacaterol (QD) has superior trough efficacy in comparison
to the two BID LABAs.formoterol9 mcg BID
indacaterol75 mcg QD
SX 50mcg BID andSpiriva 18 mcg QD
SXFP 50/250 mcg BID
86 ml
149 ml
162 ml
154 ml
95% CI: 131-167 ml
95% CI: 135-195 ml
95% CI: 136-171 ml
95% CI: 68-106 ml
Pre-SABA/SAAC
17
Several large studies indicating poor efficacy for budesonide.Due to protocol/study conduct
or compound?
Placebo-corrected FEV1 (L)
roflumilast500 mcg QD
budesonide400/320 mcg BID
Fluticasone250 mcg BID
Spiriva18 mcg QD
Respimat5 mcg QD
salmeterol50 mcg BID
0.05 0.10 0.15
36 ml
61 ml
83 ml
92 ml
128 ml
116 ml
95% CI: 117-139 ml
95% CI: 101-131 ml
95% CI: 22-51 ml
95% CI: 48-72 ml
95% CI: 71-97 ml
95% CI: 80-106 ml
Efficacy in moderate COPD95% CI including treatment heterogeneity (ISV Emax)
formoterol9 mcg BID
indacaterol75 mcg QD
SX 50mcg BID andSpiriva 18 mcg QD
SXFP 50/250 mcg BID
86 ml
149 ml
162 ml
154 ml
95% CI: 57-123 ml
95% CI: 100-196 ml
95% CI: 112-224 m
95% CI: 115-187 ml
Pre-SABA/SAAC
18Placebo-corrected FEV1 (L)
roflumilast500 mcg QD
budesonide400/320 mcg BID
Fluticasone250 mcg BID
Spiriva18 mcg QD
Respimat5 mcg QD
salmeterol50 mcg BID
0.05 0.10 0.15 0.20
36 ml
61 ml
83 ml
92 ml
128 ml
116 ml
95% CI: 20-57 ml
95% CI: 38-77 ml
95% CI: 57-113 ml
95% CI: 65-127 ml
95% CI: 87-167 ml
95% CI: 78-154 ml
Efficacy with untreated baseline 1 L95% CI including treatment heterogeneity
formoterol9 mcg BID
indacaterol75 mcg QD
SX 50mcg BID andSpiriva 18 mcg QD
SXFP 50/250 mcg BID
82 ml
142 ml
156 ml
127 ml
95% CI: 53-119 ml
95% CI: 92-191 ml
95% CI: 108-212 m
95% CI: 92-161 ml
Pre-SABA/SAAC
19Placebo-corrected FEV1 (L)
roflumilast500 mcg QD
budesonide400/320 mcg BID
Fluticasone250 mcg BID
Spiriva18 mcg QD
Respimat5 mcg QD
salmeterol50 mcg BID
0.05 0.10 0.15 0.20
23 ml
39 ml
54 ml
88 ml
122 ml
110 ml
95% CI: 13-38 ml
95% CI: 23-54 ml
95% CI: 35-76 ml
95% CI: 60-121 ml
95% CI: 81-159 ml
95% CI: 74-148 ml
Predicted efficacy: Morning trough Pre- vs. Post-SABA/SAAC
LABA and LAAC efficacy when FEV1 is measured post-
SABA/SAAC
Spiriva
Respimat5 mcg QD
salmeterol50 mcg BID
formoterol9 mcg BID
indacaterol75 mcg QD
48 ml
83 ml
52 ml
65 ml
Post-SABA/SAAC
In TORCH (ref 650) reports a disappointing 40
20
LABA and LAAC efficacy when FEV1
is measured pre-SABA/SAAC
Placebo-corrected FEV1 (L)
Spiriva18 mcg QD
Respimat5 mcg QD
salmeterol50 mcg BID
formoterol9 mcg BID
indacaterol75 mcg QD
0.05 0.10 0.15
86 ml
149 ml
92 ml
128 ml
116 ml
Pre-SABA/SAAC
Spiriva18 mcg QD 72 ml
reports a disappointing 40 ml efficacy for salmeterol.
Low efficacy due to measuring post SABA
Model Predicts LABA/LAAC Interaction LABA with LAAC background and vice versa
indacaterol75 mcg QD 95 ml
60% background
Spiriva18 mcg QD
Respimat5 mcg QD
salmeterol50 mcg BID
formoterol9 mcg BID
indacaterol75 mcg QD
34 ml59 ml
36 ml
72 ml60 ml
100% background
UPLIFT (661) reports 87-103 ml efficacy for
Spiriva. Canadian study
21Placebo-corrected FEV1 (L)
Spiriva18 mcg QD
Respimat5 mcg QD
salmeterol50 mcg BID
formoterol9 mcg BID
indacaterol75 mcg QD
0.05 0.10 0.15
86 ml149 ml
92 ml
128 ml116 ml
No DD interaction
Spiriva18 mcg QD
Respimat5 mcg QD
salmeterol50 mcg BID
formoterol9 mcg BID
75 mcg QD
55 ml59 ml
95 ml83 ml
Canadian study (686) reports 80-120 ml. 60% and
54% LABA background, respectively.
Discussion
• Placebo effect has no typical direction– Still important to account for: individual
studies have marked placebo effect• Dropout due to lack of efficacyp y
– Masks any disease progression– Reduce signs of disease modifying effects
• Low study baselinelower efficacy of anti-inflammatory drugsWeak trend towards lower efficacy in
bronchodilators (LABA or LAAC)• Mixed message in literature on individual data 22
Conclusion
• DD-interactions and covariates account for the sometimes disappointing results on FEV1 in late-stage trials
TORCH UPLIFT & rolumilast exacerbation trials– TORCH, UPLIFT & rolumilast exacerbation trials– Except LABA/LAAC interaction (UPLIFT) these do
not necessarily translate into lower effect on exacerbations!
• Model provides efficacy bench mark for published compounds, accounting for treatment heterogeneity (ISV Emax) 23