COST-EFFECTIVENESS OF GENE EXPRESSION PROFILING IN EARLY BREAST CANCER

Purpose: Gene expression profiling (GEP) is advocated as a method of risk-stratification for identifying early-stage breast cancer (ESBC) patients most likely to benefit from adjuvant chemotherapy (CTX), thereby sparing low-risk patients the toxicities and costs of CTX. Studies suggest that GEP provides more accurate predictions of cancer recurrence and CTX response than conventional clinical algorithms in ESBC. We assessed the cost-effectiveness of one common GEP test (the 21-gene signature, Oncotype DX) with two common clinical algorithms: National Comprehensive Cancer Center (NCCN) and St. Gallen guidelines for assignment of CTX in post-surgical ESBC patients.

Method: We developed probabilistic GEP decision and ESBC natural history Markov models.  The models took a health care payer's perspective over a lifetime horizon. The base case cohort was 61-year old hormone receptor-positive, lymph node-negative ESBC US patients. Data were from published literature and the Surveillance Epidemiology and End Results database. Both costs and benefits were discounted 3% annually. We assumed that all women received tamoxifen, and that those deemed high-risk received CTX, those deemed low-risk received no CTX, and 50% of those deemed intermediate-risk received CTX. We estimated incremental cost-effectiveness ratios (ICERs) for life years (LYs) and quality-adjusted LYs (QALYs). We present cost-effectiveness acceptability curves and frontiers and a value of information analysis to determine the potential value of further research in the area. Additionally, deterministic sensitivity analyses were undertaken to investigate specific patient sub-groups based on cohort ages, CTX decisions, recurrence rates, as well as the impact of the cost of the 21-gene signature.

Result: In the base case, NCCN was the least costly and the 21-gene signature was the most costly risk-stratification strategy. NCCN dominated St. Gallen guidelines in both LY and QALY analyses. The incremental cost of the 21-gene signature was $17,830/LY and $22,791/QALY vs. NCCN guidelines. ICERs increased with either increasing age or the 21-gene signature cost and decreased with either decreasing age or the 21-gene signature cost vs. NCCN (St. Gallen criteria remained dominated) in both LY and QALY analyses. When the cost of the 21-gene signature was halved, the ICERs of the 21-gene signature vs. NCCN decreased to $8,800/LY and $11,500/QALY.

Conclusion: The 21-gene signature appears to be a cost-effective alternative to conventional clinical algorithms in ESBC.