Escolar Documentos
Profissional Documentos
Cultura Documentos
www.inovation.org
Problem Statement: Specific
paucity of evidence
that CT imaging
adds value
gap: signal
processing
– signals are becoming
progressively brilliant
– analysis lags behind
Volumetric Image Analysis (VIA)
estimate value
– retrospectively
– prospectively
• small studies at a few premier centers
• options in mega, multi-national settings
value
volume
mean median max
Merck & 90%CI 95% CI
%relvar %relvar discord
partners
intra-
5.0 4.3 10.3 12.5 17.0
rater
inter-
5.3 4.4 10.8 15.7 19.1
rater
precision: Results for 7 teams
agreement between 7 teams
CV 95% CI
absolute values 25% 59%
auto-SLD VIA
overall 0.73 0.95
minimum 0.58 0.86
maximum 0.96 0.99
SLD.Average
vol.Average
80
% Survival (PR endpoint)
60
SLD.Average
vol.Average
80
% Survival (PD endpoint)
60
volumes might be
– more precise than SLDs as a basis for
RECIST
– more sensitive indicators of response
– cost effective
Therefore, GO to next step
next step: retrospective analysis Phase III
R + placebo p.o. x 14 d
A n = 123
N
D
SOC drug 1
O SOC = q 21 d
MI SOC drug 2
Z
E + MK p.o. x 14 d
n = 125
Can VIA add value?
outcome measures
– auto-SLD sum of longest diameters of all target lesions
– volume sum of corresponding tumor volumes
endpoints
– Objective Response Rate (ORR) number of subjects
with a Best Overall Response (BOR) of PR or CR
divided by the total number of subjects
– Progression Free Survival (PFS) time from first drug
dose until PD based on change in tumor mass or the
appearance of new metastases
Methods (continued)
statistical analysis
– 3D versus 1D in all subjects
– MK arm versus placebo arm
categorical variables: RECIST versus “Enhanced
RECIST”
– Partial Response (PR) = decrease of >30% from baseline
– Progressive Disease (PD) = increase of >20% from nadir
continuous variables
– median change in auto-SLD or volume at each time-point
for whole groups
Results: Objective Response Rates
Survival
allPlot
184 patients
1.0 SLD
Tests Between Groups
Volumetric
% Reaching Partial Response
1.0 SLD
0.2 Volumetric
% Reaching Partial Response
0.8
0.0
0 0.6 42 84 126 168 210 252 294 336 378
Time on Trial (days)
0.4
Kaplan-Meier
0.2 analysis shows that VIA is more sensitive than
auto-SLD
0.0 for detecting PR
0 100 200 300 400
BRR (days)
Results: Objective Response Rates
auto-SLD 23.9%
VIA 41.8%
time to Partial Response (PR)
Volumetric
0.2 0.8
0.6
0.0
0.4
0 42 84 126 168 210 252 294 336 378
0.2
Time on Trial (days)
VIA is more sensitive
0.0
than auto-SLD for patients with
longer times on trial
0 100 200
BRR (days)
300 400
Stable Disease as Best Overall Response
auto-SLD VIA conclusion
<20% before new loss for
lesions 6.0% 9.2% quants
>20% before new win for
lesions 16.8% 38.0% quants
ties between quants & no added
new lesions 10.3% 5.4% value
no new lesions AND no no
>20% 68.3% 48.9% information
extrapolation to a successful trial