RISK-BENEFIT FRAMEWORK TO EVALUATE GENE-EXPRESSION PROFILING IN EARLY STAGE BREAST CANCER: A DECISION MODEL DEVELOPED IN COLLABORATION WITH STAKEHOLDERS

Purpose: Recent comparative effectiveness research recommendations have emphasized that stakeholder involvement in research prioritization and design is critical for ultimate acceptance of results. With this recommendation in mind, an interactive risk-benefit model was developed to evaluate gene-expression profiling (GEP) to guide use of adjuvant chemotherapy in early stage breast cancer compared to clinical guidelines.

Method: A stakeholder collaborative feedback process was utilized to develop a decision model to estimate the comparative benefits and harms of GEP relative to NCCN guidelines, including disease recurrences, adverse events, and quality-adjusted life years (QALYs). The model allows users to specify the GEP, patient cohort age, adjuvant chemotherapy relative risk reduction, and adjuvant chemotherapy uptake in each risk stratum. A Markov process was utilized to estimate long-term clinical outcomes. Based on stakeholder concerns about QALY-based outcomes, alternative outcomes such as disease recurrences and adjuvant chemotherapy serious adverse events were presented. The GEPs prognostic and predictive properties were derived from published retrospective analyses of RCTs. Adjuvant chemotherapy utility decrements were derived from trial-based adverse event rates. The base case patient was a pre-menopausal woman of age 44.

Result: Preliminary analyses indicate that a 2-risk strata GEP, 3-risk strata GEP, and clinical guidelines would identify 50%, 46% and 92% of women to receive adjuvant chemotherapy, respectively. Based on this level of uptake, relative to a scenario where no patients receive adjuvant chemotherapy, these prognostic approaches would prevent 29%, 36% and 34% of distant recurrences, respectively. This scenario results in 17.4 QALYs, 18.1 QALYs, and 17.7 QALYs, respectively. These findings indicate that depending on the GEP examined, testing may lead to a net harm or net benefit relative to clinical guidelines, due to variation in risk of disease recurrence, despite avoidance of chemotherapy and side effects in 42% and 46% of women, respectively. 

Conclusion: GEP to inform adjuvant chemotherapy treatment decisions in early stage breast cancer could lead to an increase or decrease in QALYs relative to the use of clinical guidelines, depending on the GEP examined. Stakeholder feedback has indicated that acceptance of model results will likely depend on its degree of transparency, ease of use and interpretation, and ability to present relevant outcome measures. Ongoing efforts are focused on collaboration with stakeholders to align model structure and output with current needs.