A decision-analytic model was used to estimate cost-effectiveness of adopting a 21-gene assay in treatment decisions for women with early-stage N (1-3)/ER HER2-negative breast cancer.
Published Online: July 12, 2011
Burton F. Vanderlaan, MD; Michael S. Broder, MD; Eunice Y. Chang, PhD; Ruth Oratz, MD, FACP; and Tanya G. K. Bentley, PhD
Objective: To assess impact on health outcomes and healthcare expenditures of adopting a 21-gene assay for women with early-stage, minimally node-positive, estrogen receptor–positive (N (1-3)/ER ) HER2-negative breast cancer.
Study Design: We adapted a deterministic decision-analytic model to estimate costs and quality-of-life outcomes associated with chemotherapy, adverse events, supportive care, recurrence, and second primary cancers for usual care compared with care determined by the 21-gene assay recurrence score, where 71% and 54% of women, respectively, were treated with adjuvant chemotherapy. Model input data were based on national statistics, published literature, physician surveys, and Medicare Part B prices.
Methods: Annual numbers of events were multiplied by quality-adjusted life-years (QALYs) lost and costs to estimate net health and economic impacts of each strategy. Analyses were from a managed care payer perspective for the US population.
Results: Patients receiving the assay were predicted to gain 0.127 QALY and save $4359 annually from avoiding chemotherapy, adverse events, supportive care, and secondary primary tumors. For a 2-million member plan, net gains were 4.44 QALYs/year and savings were $13,476/year. Cost savings were greater for the Medicare population. Although overall results were sensitive only to reduced impact of testing and chemotherapy costs, they were still highly cost-effective (incremental cost-effectiveness ratio <$20,000/QALY).
Conclusions: Use of a 21-gene assay in patients with early-stage N (1-3)/ER HER2-negative breast cancer may improve health outcomes and add no incremental cost, thereby providing valuable insight for health plans, the Centers for Medicare & Medicaid Services, and clinicians regarding coverage policies and treatment decisions.
(Am J Manag Care. 2011;17(7):455-464)
By providing better predictions of recurrence risk, a 21-gene assay in patients with earlystage N (1-3)/ER HER2-negative breast cancer has the potential to significantly improve US healthcare quality at no additional payer cost.
Our model predicted that the substantial savings in chemotherapy-related costs would outweigh the cost of the test itself and that patients would experience substantial quality-oflife gains associated with reductions in chemotherapy.
These results provide insight for health plans, the Centers for Medicare & Medicaid Services, and clinicians when establishing coverage policies and making treatment decisions.
Large clinical trials have demonstrated significant benefits of combining chemotherapy with hormonal therapy in the adjuvant treatment of women with node-negative, estrogen receptor– positive (N−/ER ) breast cancer.1,2 Hormonal treatment with tamoxifen alone after surgery in this population has been shown to reduce the 10-year likelihood of distant recurrence to 15%.1 Depending on disease and tumor characteristics, adding chemotherapy can further reduce recurrence risk in some N−/ER patients, although others derive no added benefit from adjuvant chemotherapy.3,4
The 21-gene Oncotype DX Recurrence Score reverse transcriptase polymerase chain reaction assay can reliably predict individual recurrence risks among N−/ER patients based on gene expression in the tumor tissue.5-7 The assay has been incorporated into National Comprehensive Cancer Network (NCCN) and American Society of Clinical Oncology (ASCO) treatment guidelines for N−/ER patients with HER2-negative tumors8,9 and has allowed oncologists to move from population-based to individualized estimates of recurrence risk, reducing unnecessary chemotherapy treatment among patients with low predicted recurrence risk.10-14
In node-positive, estrogen receptor–positive (N (1-3)/ER ) breast cancer, however, NCCN and ASCO guidelines recommend treatment with both hormonal therapy and chemotherapy for most HER2- negative cancers, although recent research suggests that up to 40% of these patients may remain disease-free without chemotherapy.8,9,15 In a recent analysis of the TransATAC study, results of the 21-gene assay were significantly associated with time to distant recurrence and provided significant prognostic value beyond that provided by Adjuvant! Online in both N− and N (1-3)/ER cancers.7 In addition, an analysis of the SWOG-8814 clinical trial confirmed the lack of benefit of chemotherapy in women with N (1-3)/ER disease who had low recurrence scores (<18).15
Better predictions of recurrence risk could help oncologists individualize treatment recommendations. Eliminating chemotherapy for patients unlikely to benefit from it might both decrease costs and improve quality of life.11,16-18 We built a decision-analytic model to predict cost-effectiveness and cost savings of using the 21-gene assay Oncotype DX in women with earlystage N (1-3)(N1a-N1c)/ER HER2-negative breast cancer; the cost-effectiveness and cost savings were compared with the cost of treating this early-stage breast cancer according to current US guidelines.
We adapted a deterministic decision-analytic model (Figure) to estimate the cost-effectiveness of adopting the 21-gene assay in treatment decisions for women with N (1-3)/ ER HER2-negative early-stage breast cancer, comparing outcomes for 2 scenarios: usual care, in which chemotherapy treatment decisions reflected historical standards based on US NCCN guidelines; and genomic assay testing, in which decisions were modified based on assay results. The model followed women from being disease free to living with nonprogressed and progressed disease, and then to death either from disease or other causes. Model inputs were obtained from published literature, national statistics, randomized clinical trials, Medicare part B prices, and surveys of breast cancer patients and medical oncologists.
Patients in the model faced risks of being diagnosed with N (1-3)/ER HER2-negative breast cancer with varying recurrence risks; being tested with the 21-gene assay; receiving chemotherapy; experiencing adverse events, second primary cancers, or distal recurrence; and dying within a 30-year time horizon (Table 1). Outcomes of interest were incremental quality-adjusted life-years (QALYs, per patient and total plan), costs (per patient, per member per month, and total plan), and incremental cost-effectiveness associated with the use of the assay in adjuvant chemotherapy treatment decisions. Patients in both usual care and testing strategies either received chemotherapy followed by hormonal therapy or hormonal therapy alone. The single difference between usual care and assay testing strategies was that fewer patients in the testing group received chemotherapy, which in turn affected costs and quality-of-life decrements associated with chemotherapy and related adverse events, supportive care, and second primary cancers.
We evaluated the model for a hypothetical cohort of 2 million total health plan members with an age distributionrepresentative of the US population, estimating results for all women with N (1-3)/ER HER2-negative breast cancer. We also evaluated the model for a subset of those 65 years and older. We did not address assay use among patients with N- or HER2 tumors. Analyses were from a payer (managed care) perspective, and annual risks were based on average risks over a 30-year time horizon. Models were developed and analyzed using Microsoft Office Excel 2003 (Redmond, WA).
Table 1 shows parameter estimates used in the deterministic decision-analytic model. We used 2004-2005 SEER*STAT data to estimate an age-adjusted incidence of early-stage N (1-3)/ER breast cancer of 9.74 diagnoses per 100,000 population and a 90% prevalence of HER2-negative tumors, using an age distribution representative of the US population and a mean age of diagnosis of 62 years.19 We considered women to be N if they had 1-3 positive nodes based on evidence that most N patients receiving the assay had <3 positive nodes.21 Diagnosed patients in usual care were assumed to be treated according to national guidelines that recommended adjuvant chemotherapy treatment for all N (1-3)/ ER patients.8 Because some patients may refuse chemotherapy and oncologists may not recommend chemotherapy in all cases (eg, among N1a patients with only 1 positive node), we assumed that 71% of women in usual care would receive chemotherapy treatment.20
In the testing strategy, we assumed that 20% of diagnosed patients would be tested with the assay. Among these patients, we estimated a 24% relative reduction in chemotherapy use associated with assay results based on data from a survey of oncologists who used the assay in this patient population.21 Compared with the 71% uptake in usual care, 54% [(1.0 – 0.24 x 71%)] of tested women would receive chemotherapy treatment. A post hoc analysis of the same survey showed a 31% relative reduction in chemotherapy for patients 65 years and older; we thus assumed a 49% chemotherapy rate [(1.0 – 0.31) x 71%)] among a subset of tested women in this
age group.26 We assumed that the decrease in chemotherapy associated with assay testing occurred only in patients with low recurrent score,7,15 so disease recurrence rates did not differ between strategies. Age-specific all-cause mortality was estimated based on national statistics, and for age- and disease-specific mortality we used SEER estimates of overall agespecific breast cancer mortality.19,39 These estimates were used to calculate total QALYs lost with each strategy.
We categorized chemotherapy-related adverse events into minor events, major events, fatal events, and second primary cancers. Minor adverse events were defined as grade <2 (mild or moderate), where applicable, of the following: chemotherapy- induced nausea and vomiting; neutropenia; thrombocytopenia; diarrhea; alopecia; cardiovascular-functional (defined as any cardiovascular event that was asymptomatic, transient, or responded to treatment); phlebitis; infection; hemorrhagic cystitis; fever; and weight gain or loss. When these adverse events were grade >3 (severe, life threatening, or fatal), they were considered major, as were neurosensory or neuromotor toxicity, arthralgia, myalgia, granulocytopenia, hypersensitivity reaction, thromboembolic event, ovarian failure, cognitive dysfunction, congestive heart failure, and febrile neutropenia. Any such minor or major events that caused death were considered separately as fatal events.
The incidence of minor adverse events was based on published reports of key trials of anthracycline therapy and antiemetic regimens.7,15,28-30,32,35-37 We estimated an 85% minor adverse event rate to account for the higher rates of chemotherapy-induced nausea and vomiting seen in patients receiving some newer regimens and the almost-universal alopecia occurrence seen in patients receiving other regimens.35,40 We estimated the probability of major adverse events based on published literature indicating that risk of these events among N breast cancer patients ranges between 20% and 35%,33,34 and we assumed an average incidence of 30%.
We estimated a 2.7% probability of second primary cancer based on the assumption of a 20-year survival time among diseased patients without recurrence or second primary cancer and a 13.6/10,000 person-years absolute excess risk of second primary nonbreast cancer after treatment (20 x 13.6/10,000 = 2.72%).34,38 Use patterns of chemotherapyrelated supportive care (including treatment with granulocyte colony-stimulating factor, granulocyte-macrophage colony-stimulating factor, erythropoietin, and antiemetics) were based on a 2007 survey of medical oncologists in private oncology practices.27
PDF is available on the last page.