cost-effectiveness fdg-pet in lung cancer staging
TRANSCRIPT
Cost-Effectiveness
FDG-PET in
Lung Cancer Staging
Small cell cancer 20~25% Squamous ca 30~35% Adenocarcinoma 30~35% Large cell ca 15~20%
BackgroundEpidemiology of Lung Cancer
15% of all cancer in USA Yearly new cases: 3 mill. wordwide, 0.18 mill. USA
NSCLC
Single most important prognostic factor in NSCLC = resectability Resectable in 20~35% of NSCLC patients.
결핵 및 호흡기 질환 1999
Staging of NSCLC
IA T1N0M0 67 +
IB T2N0M0 57 +
IIA T1N1M0 55 +
IIB T2N1M0
T3N0M0
IIIA T3N1M0
T1~3N2M0
IIIB T1~4N3M0
T4N0~2M0
IV any M1 1 –
Stage Definition 5Y surv(%) Resectability
38~39
23~25
1~8
+
N3
N2
++
Peribronchial LN
Ipsilateral hilar LN
Intrapulmonary LN
N1
Ipsilateral mediastinal LN
Subcarinal LN
N2
Contralateral mediastinal
Contralateral hilar LN
Supraclavicular LN
N3
CT in NSCLC Staging
CT is the standard method for screening for LN involvement
Meta-analysis of 42 studies: 79% SN, 78% SP More recent reports : 52-64% SN, 62-69% SP
CT accuracy for LN assessment
Variation in normal LN size range Reactive hyperplasia or infection related enlargement LN not enlarged with microsccopic metastasis Poor inter-observer kappa value for size assessment
Limitations of LN assessment with CT
FDG PET in NSCLC Staging
Meta-analysis : 87% SN, 95% SP
PET accuracy for LN assessment
Detect metastasis in non-enlarged LNs Exclude metastasis in enlarged LNs
Potential role of PET in LN staging
FDG PET Staging in ROK
Total N = 20 Biopsy N = 17 N0 11; N1 2; N2 4 CT : SN 50%; SP 47% PET: SN 83%; SP 73%
Total N = 25 Biopsy N = 25 LN stage ? CT : accuracy 56% PET: accuracy 76%
SNUH, FDG PET, 1997
Ajou U, CoDe PET, 1999
Evidence Profile for PET
Diagnosticaccuracy
Technical performance
Healthimpact
Therapeuticimpact
Diagnosticimpact
Majority
Costeffectiveness
Eur J Rad, 1999
limited number a few (modeling)
1) Decision-tree model construction
Decision Tree Analysis
Gambhir et al. JNM 1996
2) Medical literature survey
3) Calculation of expected cost & effectiveness
4) Sensitivity analysis
• evaluate each tree over variable range
• determine break-even point for variables
Decision Tree Model
StrategiesBx (-)
Surgery (+) Nodes
Surgery (-) Nodes
CT (+)
Bx (+)Op
Bx
CT
Surgery (+) Nodes
Surgery (-) Nodes
CT (-)
Op
PET
PET (-)
CT (-)
Surgery (-) Nodes
Surgery (+) Nodes
CT
Op
CT-only
PET+CT Surgery (+) Nodes
Surgery (-) Nodes
CT (+)
BxBx (+)
Bx (-) Op
Surgery (-) Nodes
Surgery (+) Nodes
Bx (+)PET (+)
Bx (-)
BxOp
CT
Unresectable rate 31 %
CT 예민도 67 %
CT 특이도 73 %
PET 예민도 90 %
PET 특이도 91 %
생검 morbitity 0.
007 yr
수술 morbitity 0.083 yr
PET 사망율 0.0000 % CT 사망율 0.0025 %
생검 사망율 0.3 %
수술 사망율 3.0 %
LE-resectable 7 yr
LE-unresectable 1 yr
Medical Literature Survey
Effectiveness = life expectancy (LE)
Calculation of Expected Cost & Effectiveness
Cost = medical expenditure
• CT cost = 700 $• PET cost = 1,000 $• Biopsy cost = 3,000 $• Surgery cost = 30,000 $
Declining exponential approximation of LE method
annual mortality of gen pop. + by dsLE =
1
( )
RESULTS
• Average 1,154 $ savings per patient (25,634$ vs. 24,480$)
• No loss of LE (2.96 day gain)
Conservative CT+PET Strategy (A)
• Average 2,267 $ savings per patient (25,634 $ vs. 23,367 $)
• Failure to operate on 1.7% of resectable cases
Less conservative CT+PET Strategy (C)
• Average 1,154 $ savings per patient (25,634$ vs. 24,480$)
• No loss of LE (2.96 day gain)
Conservative CT+PET Strategy (A)
26.2
25.8
25.4
25.0
24.6
24.2
0.300 0.475 0.650 0.825 1.000
PET Sensitivity
Exp
ecte
d V
alue
(th
ousa
nd $
)
CT
CT + PET
Threshold Values SENS = 0.48 EV = $25,600
Sensitivity Analysis
Threshold Values
SPEC = 0.12
EV = $25,600
25.7
25.5
25.3
25.1
24.9
24.7
24.5
24.3
0.100 0.325 0.550 0.775 1.000
PET Specificity
Exp
ecte
d V
alue
(th
ousa
nd $
)
CT CT + PET
Sensitivity Analysis
1.00
0.94
0.88
0.82
0.76
0.700.15 0.19 0.23 0.27 0.31 0.35
Prevalance
PE
T S
pec
ific
ity
CT
CT + PET
Two-way sensitivity analysis
5,000
4,100
3,200
2,300
1,400
5005,000 14,000 23,000 32,000 41,000 50,000
Surgery Cost ($)
PE
T C
ost
($)
CT
CT + PET
Two-way sensitivity analysis
Scott et al. Ann Thorac Surg, 1998
Variables different from 1st study
PET 예민도 90 %
PET 특이도 91 %
PET 예민도 , CT+ 89 %PET 예민도 . CT- 76 %
PET 특이도 , CT+ 81 % PET 특이도 , CT- 97 %
CT cost 700 $ PET cost 1,000 $ Biopsy cost 3,000 $ Surgery cost 30,000 $
378 $ 2,000 $ 4,360 $18,500 $
Decision Tree Model
Strategies
CT alone
Bx (-)
Surgery (+) Nodes
Surgery (-) Nodes
CT (+)
Bx (+)
Surgery (+) Nodes
Surgery (-) Nodes
CT (-)
CT
Bx
Op
Op
CT
CT (-)
PET (-)
Surgery (-) Nodes
Surgery (+) Nodes
PET
Op
Surgery (+) Nodes
Surgery (-) Nodes
PET (+)
BxBx (+)
Bx (-) Op
Surgery (-) Nodes
Surgery (+) Nodes
Bx (+)CT (+)
Bx (-)
BxOp
PET
(CT+PET)
Strategies
CT+PET *
CT (-) PET
PET (-)Surgery (-) Nodes
Surgery (+) Nodes
Op
Surgery (+) Nodes
Surgery (-) Nodes
PET (+)
Op
Bx (+)
Bx (-)
Surgery (-) Nodes
Surgery (+) Nodes
Bx (+)
CT (+)
Bx (-)
CT
Bx
Op
RESULTS
Strategy CT alone (CT+PET) CT+PET*
Cost 16,743 $ 17,708 $ 16,920 $
LE 4.921 y 4.928 y 4.928 y
ICER ( - ) 137,857 25,286
* Incremental cost-effectiveness ratio (ICER)
(LEX - LE A)ICER =
(CostX-CostA)
100
80
60
40
20
0
-20
-40
1400 1600 1800 2000 2200
Cost PET ($)
ICE
R (
$1,0
00/Y
r)
BCD
Sensitivity Analysis
100
80
60
40
20
0
-20
-40
0.25 0.3 0.35 0.4
Prevalence
ICE
R (
$1,0
00/Y
r)
BCD
Sensitivity Analysis
PET distant metastasis of NSCLC
Lewis (1994)
Bury(1996)
Rigo(1997)
Valk(1996)
Tx plan change in 41%Surgery avoided in 18%
N-stage change in 21%M-stage change in 10%
M stage change in 14%
Metastasis detected in 11%Confirm metastasis in 7%Exclude metastasis in 16%
n = 34
n = 61
n = 39
n = 99
• PET 700 $
• Surgery, benign 8,323 $
• Surgery, malign 16,377 $
FDG PET Cost-Effectiveness in Japan
Analysis Method
Cost
• 7.3 mo LE increase effect
• ICER = 20 만 Yen / yr.pt
Results of CT+PET strategy
• Decision-tree analysis
Kosuda et al. Chest, 2000
FDG PET Cost-Effectiveness in ROK ?
Unresectable rate 31 %
CT 예민도 67 %
CT 특이도 73 %
PET 예민도 90 %
PET 특이도 91 %
수술 사망율 3.0 %
CT cost
PET cost
Biopsy cost
Surgery cost
700 $
1,000 $
3,000 $
30,000 $
Variables and Cost ?
Management policies ?
Health Care Environment ?
SNU 83%73%
Ajou
76%