Are Further Studies of Breast Cancer Tumor Markers Justified? A Value of Research AnalysisRahber Thariani PhD1, David Blough PhD1 , Norah Lynn Henry MD, PhD2, Bill Barlow PhD1, Julie Gralow MD13, Scott Ramsey MD PhD13, David Veenstra PharmD PhD1

1University of Washington, 2University of Michigan, 3Fred Hutchinson Cancer Research Center,

Center for Comparative Effectiveness Research in Cancer Genomics (CANCERGEN)OBJECTIVE

INTRODUCTION

BREAST CANCER MODEL

CONCLUSIONS

Expected Value of Sample Information (EVSI)

EVSI - CONFIDENCE IN EXPERT OPINIONAND TRIAL SIZE

RESULTS

KEY PARAMETER VALUESParameter

Mean Value

CI 95% range

Distribution

Early detection and treatment (HR)

1.15 0.9-1.4 Normal

Probability of recurrence in 5 year

interval24.5% 23%-26% Normal

Tumor Marker Sensitivity

57.7% 37.7%-77.7% Beta

Tumor Marker Specificity

97.9% 96%-99% Beta

Utility – No Disease 0.92 0.8-1 BetaUtility standard chemotherapy

0.62 0.49-0.75 Normal

Incremental utility – early detection

and chemotherapy0.05 0-0.1 Beta

Disutility False positive test result (1 year duration)

0.19 0.13-0.25 Normal

To assess the value of additional research for testing carci-noembroynic antigen (CEA), cancer antigen (CA)15-3 and CA 27.29 biomarkers for earlier detection and treatment of recurrent breast cancer

In a recent trial a CA 27.29 radioimmunoassay was able to identify patients with breast cancer recurrence 5.3 months before recurrence is clinically established. However, Ameri-can Society of Clinical Oncology (ASCO) guidelines recom-mend against the serial use of these tumor markers to detect recurrence following breast cancer therapy. How-ever, usage estimates for biomarker testing exceed 20% of all cases of early breast cancer.The ASCO recommendation is based on the absence of data showing a survival or other outcome benefits such quality of life, drug toxicity etc.Two prospective trials con-ducted in the 80s following breast cancer patients with in-tensive vs. standard follow-up regime showed no significant differences in overall survival. Given newer therapies with improved efficacies and toxicity profiles, earlier detection and treatment of breast cancer recurrence may yield sub-stantial improvements in healthcare outcomes. However, a trial would necessary to determine if breast cancer tumor markers should be actively used in standard of care breast cancer recurrence surveillance.Value of research methods can be used to evaluate if such a trial would represent an efficient use of resources.

IS THE TRIAL WORTH IT?WHAT INFORMATION WILL HAVE THE MOST IMPACT?

HOW DO THE BENEFITS COMPARE WITH OTHER PROPOSED TRIALS?

A decision-analytic model was developed with biomarker testing in addition to standard surveillance at follow-up ap-pointments for breast cancer patients every 3 months for five years, following completion of primary therapy. Esti-mates of survival and survival uncertainty were obtained through expert opinion, due to lack of clinical data. Other uncertainties and model parameters, including quality-of-life indicators associated with testing, were derived from litera-ture values. The affected population was estimated from SEER incidence data and discounted over a 10-year time horizon. We assumed that the 5-year recurrence rate was 25% and all recurrence cases were metastatic in nature.

Our results indicate that substantial value to society can be obtained by evaluating the clinical utility of serial tumor marker assessment for early detection of breast cancer re-currence. This value is driven by the current paucity and conflicting information in this area, severity of outcomes, and large population that could be affected. Even small trials focusing on reducing uncertainty in specific param-eters, may provide substantial value.

An analysis was also conducted to determine the impact of different trial sizes. We also explored the imapact of dif-ferent levels of confidence in expert opinion i.e. being equivalent to an existing trial with n=1 to n=50. Note that the sharp edges below are due to convergence errors.

An emerging field in health economics—value of informa-tion (VOI) analysis—quantifies the value of future research, and may be helpful in resource allocation decisions. Based on Bayesian decision theory, these methods provide an analytical framework to assess the societal value obtained by the reduction of uncertainty around a treatment or testing decision. The value of information is a function of 1) the probability of selecting the optimal monitoring strategy based on current vs. future information, 2) the clinical and economic impacts of each strategy, and 3) the size of the population affected.

Early Stage Breast Cancer

Patients following Primary Therapy

Biomarker Testing + Standard

Surveillance

2x Elevated Biomarker

Levels

Confirmatory Test (Bone Scan,

CT, PET)

Early Chemotherapy (if confirmed)

Non-Elevated Biomarker

Levels

Standard Chemotherapy if

recurrence

Standard Surveillance

Standard Chemotherapy if

recurrence