Methods to Analyse The Economic Benefits of a Pharmacogenetic (PGt) Test to Predict Response to Biologic Therapy in Rheumatoid Arthritis, and to Prioritise Further Research Alan Brennan 1, Nick Bansback 1, 1 ScHARR, University of Sheffield, England. Kip Martha 2, Marissa Peacock 2, Kenneth Huttner 2 2 Interleukin Genetics, Inc.
“Biologics” Anakinra ($12,697), Etanercept ($18,850), Infliximab ($24,112)* * Costs include monitoring Anakinra 100mg Etanercept 25mg eow Infliximab 3mg/kg 8 weekly
“Biologics” Anakinra ($12,697), Etanercept ($18,850), Infliximab ($24,112)* Cytokines Interleukin 1TNF alpha TNF Alpha * Costs include monitoring Anakinra 100mg Etanercept 25mg eow Infliximab 3mg/kg 8 weekly
“Biologics” Anakinra ($12,697), Etanercept ($18,850), Infliximab ($24,112)* Is Response Genetic? 91 patients, 150mg Anakinra, 24 week RCT 1,2, gene = IL-1A Positive response = reduction of at least 50% in swollen joints 1 Camp et al. American Human Genetics Conf abstract 1088, Bresnihan Arthritis & Rheumatism, 1998 * Costs include monitoring Anakinra 100mg Etanercept 25mg eow Infliximab 3mg/kg 8 weekly
“Biologics” Anakinra ($12,697), Etanercept ($18,850), Infliximab ($24,112)* Is Response Genetic? 24 week RCT 1,2, 91 patients, 150mg Anakinra,, gene = IL-1A Defined response = reduction of at least 50% in swollen joints 1 Camp et al. American Human Genetics Conf abstract 1088, Bresnihan Arthritis & Rheumatism, 1998 * Costs include monitoring Anakinra 100mg Etanercept 25mg eow Infliximab 3mg/kg 8 weekly
“Biologics” Anakinra ($12,697), Etanercept ($18,850), Infliximab ($24,112)* Is Response Genetic? 91 patients, 150mg Anakinra, 24 week RCT 1,2, gene = IL-1A Positive response = reduction of at least 50% in swollen joints 1 Camp et al. American Human Genetics Conf abstract 1088, Bresnihan Arthritis & Rheumatism, 1998 * Costs include monitoring Anakinra 100mg Etanercept 25mg eow Infliximab 3mg/kg 8 weekly 50% 100%
Health Outcomes ACR20 response -20% in swollen, and tender joints, and in 3 other measures
Health Outcomes ACR20 response -20% in swollen, and tender joints, and in 3 other measures ACR20 = 0.88 * Swollen50 score (trial data)
Health Outcomes ACR20 response -20% in swollen, and tender joints, and in 3 other measures ACR20 = 0.88 * Swollen50 score (trial data) Response ==> symptom relief and delayed progression long term
Health Outcomes ACR20 response -20% in swollen, and tender joints, and in 3 other measures ACR20 = 0.88 * Swollen50 score (trial data) Response ==> symptom relief and delayed progression long term “Years in ACR20 Response” = primary outcome 3 Kobelt et al. Economic Conseque of Progression of RA in Swe. A&R 1999
Health Outcomes ACR20 response -20% in swollen, and tender joints, and in 3 other measures ACR20 = 0.88 * Swollen50 score (trial data) Response ==> symptom relief and delayed progression long term “Years in ACR20 Response” = primary outcome ACR 20 Response 0.8 reduction in HAQ (0 to 3 scale) Utility * HAQ 3 3 Kobelt et al. Economic Conseque of Progression of RA in Swe. A&R 1999
Existing Uncertainty 50%
2 Year Treatment Sequence Pathway Initial Response Longer term discontinuation
A Pharmaco-Genetic Strategy Strategy 1 Strategy 2
Strategy Sequences to Compare A Anakinra PGtGenetic EEtanercept IInfliximab - Maintenance
Existing Uncertainty (2)
Cost Assumptions Drugs and Monitoring Other Healthcare HAQ $Cost pa = $1,084 + $1,636 * HAQ 4 ==> Responder = $ 2,400 pa Non Responder = $ 3,700 pa PGt = $200 Excluding :Nursing Home Care, Employer Costs No uncertainty analysis 4 Yelin and Wanke. A&R 1999………...
2 Level EVSI - Research Design 4, 5 4 Brennan et al Poster SMDM Brennan et al Poster SMDM 2002
2 Level EVSI - Research Design 4, 5 4 Brennan et al Poster SMDM Brennan et al Poster SMDM )Decision model, threshold, priors for uncertain parameters
2 Level EVSI - Research Design 4, 5 4 Brennan et al Poster SMDM Brennan et al Poster SMDM )Decision model, threshold, priors for uncertain parameters 1) Simulate data collection:
2 Level EVSI - Research Design 4, 5 4 Brennan et al Poster SMDM Brennan et al Poster SMDM )Decision model, threshold, priors for uncertain parameters 1) Simulate data collection: sample parameter(s) of interest once ~ prior (1st level)
2 Level EVSI - Research Design 4, 5 4 Brennan et al Poster SMDM Brennan et al Poster SMDM )Decision model, threshold, priors for uncertain parameters 1) Simulate data collection: sample parameter(s) of interest once ~ prior decide on sample size (n i ) (1st level)
2 Level EVSI - Research Design 4, 5 4 Brennan et al Poster SMDM Brennan et al Poster SMDM )Decision model, threshold, priors for uncertain parameters 1) Simulate data collection: sample parameter(s) of interest once ~ prior decide on sample size (n i ) (1st level) sample a mean value for the simulated data | parameter of interest
2 Level EVSI - Research Design 4, 5 4 Brennan et al Poster SMDM Brennan et al Poster SMDM )Decision model, threshold, priors for uncertain parameters 1) Simulate data collection: sample parameter(s) of interest once ~ prior decide on sample size (n i ) (1st level) sample a mean value for the simulated data | parameter of interest
2 Level EVSI - Research Design 4, 5 4 Brennan et al Poster SMDM Brennan et al Poster SMDM )Decision model, threshold, priors for uncertain parameters 1) Simulate data collection: sample parameter(s) of interest once ~ prior decide on sample size (n i ) (1st level) sample a mean value for the simulated data | parameter of interest 2) combine prior + simulated data --> simulated posterior
2 Level EVSI - Research Design 4, 5 4 Brennan et al Poster SMDM Brennan et al Poster SMDM )Decision model, threshold, priors for uncertain parameters 1) Simulate data collection: sample parameter(s) of interest once ~ prior decide on sample size (n i ) (1st level) sample a mean value for the simulated data | parameter of interest 2) combine prior + simulated data --> simulated posterior 3) now simulate 1000 times parameters of interest ~ simulated posterior unknown parameters ~ prior uncertainty (2 nd level)
2 Level EVSI - Research Design 4, 5 4 Brennan et al Poster SMDM Brennan et al Poster SMDM )Decision model, threshold, priors for uncertain parameters 1) Simulate data collection: sample parameter(s) of interest once ~ prior decide on sample size (n i ) (1st level) sample a mean value for the simulated data | parameter of interest 2) combine prior + simulated data --> simulated posterior 3) now simulate 1000 times parameters of interest ~ simulated posterior unknown parameters ~ prior uncertainty (2 nd level) 4) calculate best strategy = highest mean net benefit
2 Level EVSI - Research Design 4, 5 4 Brennan et al Poster SMDM Brennan et al Poster SMDM )Decision model, threshold, priors for uncertain parameters 1) Simulate data collection: sample parameter(s) of interest once ~ prior decide on sample size (n i ) (1st level) sample a mean value for the simulated data | parameter of interest 2) combine prior + simulated data --> simulated posterior 3) now simulate 1000 times parameters of interest ~ simulated posterior unknown parameters ~ prior uncertainty (2 nd level) 4) calculate best strategy = highest mean net benefit 5) Loop 1 to 4 say 1,000 times Calculate average net benefits
2 Level EVSI - Research Design 4, 5 4 Brennan et al Poster SMDM Brennan et al Poster SMDM )Decision model, threshold, priors for uncertain parameters 1) Simulate data collection: sample parameter(s) of interest once ~ prior decide on sample size (n i ) (1st level) sample a mean value for the simulated data | parameter of interest 2) combine prior + simulated data --> simulated posterior 3) now simulate 1000 times parameters of interest ~ simulated posterior unknown parameters ~ prior uncertainty (2 nd level) 4) calculate best strategy = highest mean net benefit 5) Loop 1 to 4 say 1,000 times Calculate average net benefits 6) EVSI parameter set = (5) - (mean net benefit | current information)
2 Level EVSI - Research Design 4, 5 4 Brennan et al Poster SMDM Brennan et al Poster SMDM )Decision model, threshold, priors for uncertain parameters 1) Simulate data collection: sample parameter(s) of interest once ~ prior decide on sample size (n i ) (1st level) sample a mean value for the simulated data | parameter of interest 2) combine prior + simulated data --> simulated posterior 3) now simulate 1000 times parameters of interest ~ simulated posterior unknown parameters ~ prior uncertainty (2 nd level) 4) calculate best strategy = highest mean net benefit 5) Loop 1 to 4 say 1,000 times Calculate average net benefits 6) EVSI parameter set = (5) - (mean net benefit | current information)
4 strategies: A, E, I and PGt Results - 6 months
4 strategies: A, E, I and PGt Results - 6 months
4 strategies: A, E, I and PGt Results - 6 months
20 strategies: A, E, I and PGt sequences Base-case Results - 2 years
20 strategies: A, E, I and PGt sequences Optimal Strategy Depends on Threshold: $10k ==> maintenance therapy(20) $20k ==> sequence of 2 biologics(11) $25k ==> PGt + 2 biologics (9) $30k ==> PGt + 3 biologics(19) Base-case Results - 2 years
20 strategies: A, E, I and PGt sequences Optimal StrategyProb Depends on Threshold: Optimal $10k ==> maintenance therapy(20)100% $20k ==> sequence of 2 biologics(11) 42% $25k ==> PGt + 2 biologics (9) 18% $30k ==> PGt + 3 biologics(19) 43% Base-case Results - 2 years
Incorporating Uncertainty Assuming 25,000 per annum new patients starting biologics over next 5 years
Partial EVPI: Key Uncertainties
Partial EVSI: PGt Research only Caveat: Small No.of Simulations on 1st Level
Interleukin Genetics Inc. TARGET RA program Conceptual modelling identified key missing data and helped prioritise further primary data collection 1. PGt test performance (increased sample size). 2. Etanercept / Infliximab performance in gene subgroups 3. Anakinra response rate in anti-TNF α failures
Partial EVPI: TARGET RA Program
Conclusions Early economic evaluation suggests potential for a cost-effective pharmacogenetic test.
Conclusions Early economic evaluation suggests potential for a cost-effective pharmacogenetic test. Expected value of information analysis has quantified the key research priorities.
Conclusions Early economic evaluation suggests potential for a cost-effective pharmacogenetic test. Expected value of information analysis has quantified the key research priorities. EVSI can quantify the value of the specific research design