Oncotype DX breast cancer recurrence score test: A single technology assessment
Health technology assessment
|Published
The objective of the STA was to appraise the evidence addressing the following questions: (Q1) Can Oncotype DX predict chemotherapy benefit? (Q2) Does Oncotype DX provide prognostic information? (Q3) What is the distribution of RS in populations of breast cancer patients? (Q4) Can Oncotype DX reduce chemotherapy use?
Summary
Introduction
Early-stage breast cancer patients undergo surgery to remove the primary tumor. After surgery, patients characterized as estrogen reseptor positive (ER+) and human epidermal growth factor receptor 2 negative (HER2-) are routinely treated with endocrine therapy with or without chemotherapy to prevent recurrence. Chemotherapy causes side-effects and should ideally only be offered to patients who benefit from the treatment.
In this single technology assessment (STA), we have considered the gene-profiling test Oncotype DX. The test calculates a Recurrence Score (RS) between 0 and 100, and is intended to estimate the risk of recurrence, and to predict whether breast cancer patients will benefit from chemotherapy. Our assessment is based on documentation submitted by Oecona on behalf of the manufacturer of Oncotype DX, Exact Sciences.
Objective
The objective of the STA was to appraise the evidence addressing the following questions: (Q1) Can Oncotype DX predict chemotherapy benefit? (Q2) Does Oncotype DX provide prognostic information? (Q3) What is the distribution of RS in populations of breast cancer patients? (Q4) Can Oncotype DX reduce chemotherapy use?
We also appraised the cost-effectiveness and budget impact analyses provided by the submitter.
Method
Clinical effectiveness: We performed a separate literature search to evaluate whether all randomized controlled trials (RCTs) addressing question Q1 were found and included by the submitter. We extracted data from the RCTs and critically appraised the risk of bias. Our confidence in the results was assessed using the GRADE approach. We also extracted data from RCTs and non-randomized studies to validate the submitted evidence for question Q2, Q3, and Q4.
Health economics: The health economic model provided by the submitter combined a decision tree and a Markov model to compare the Oncotype DX test strategy to assessment of traditional clinical parameters (no gene-profiling test). Patients were categorised as having low, intermediate or high RS, and in each RS group, patients received or did not receive chemotherapy. These groups were connected to a Markov model that predicted lifetime QALYs and costs, considering the risk of distant recurrence. The submitter also performed sensitivity analyses and budget impact analyses.
Results
Clinical effectiveness: The submitter identified four RCTs that investigated whether Oncotype DX can predict chemotherapy benefit in patients with ER+ HER- early-stage breast cancer. Two of the RCTs investigated patients with node negative disease and two investigated node positive disease (1-3 positive lymph nodes). Additional RCTs were not identified in our separate literature search.
For patients with node negative disease (regardless of menopausal status) and for postmenopausal patients with node positive disease, there was convincing evidence that:
- Patients with low or intermediate RS (0-25) have similar risk of recurrence regardless of whether they are treated with endocrine therapy plus chemotherapy or endocrine therapy alone (have no chemotherapy benefit). However, a small chemotherapy benefit was observed for node negative women below 50 years and a RS of 16-25.
- Patients with high RS (>25) treated with endocrine therapy plus chemotherapy have lower risk of recurrence than those treated with endocrine therapy alone (have chemotherapy benefit).
RCTs and non-randomized studies also demonstrated that Oncotype DX provides prognostic information. Distributions of RS suggested that approximately 75-90% of the patients with node negative and node positive disease have low or intermediate RS (0-25) and can omit chemotherapy. Decision-impact studies demonstrated that Oncotype DX reduced chemotherapy assignment by 16-27% for node negative patients and by 50-73% for node positive patients. The studies also illustrated that treatment decisions were not based entirely on the RS.
Health economics: The base case cost-effectiveness analysis indicated that Oncotype DX was dominant compared to assessment of traditional clinical parameters (no gene-profiling test) for both node negative and postmenopausal node positive patients (i.e., provided greater QALY gains at a lower cost). One way sensitivity analyses identified the primary sources of uncertainty in the model. These were the hazard ratio for the high RS group for node negative patients, and the hazard ratio for the low and intermediate RS group for postmenopausal node positive patients. Probabilistic sensitivity analyses showed that, with a willingness-to-pay threshold of 250,000 NOK per QALY, the Oncotype DX test had a probability of 99% and 100 % of being cost-effective (compared to no gene profiling test) for node negative and postmenopausal node positive patients respectively. Absolute shortfall of QALYs was calculated to be 1.83 QALYs.
Budget impact analyses showed net costs of implementing Oncotype DX (compared to no gene-profiling test) for node negative patients, and net savings for postmenopausal node positive patients, each of the five years following implementation. The budget impact analysis for the entire population showed net costs each of the five years. The analyses are associated with uncertainty.
Discussion
Four RCTs demonstrated that Oncotype DX predicted chemotherapy benefit for node negative and postmenopausal node positive breast cancer patients. The RCTs were well designed and risks of bias were deemed low. Together the four RCTs provided convincing evidence, and this represent a major strength.
The gene profiling test Prosigna was approved for node negative patients by the Norwegian Decision Forum in 2019, and is now implemented in clinical practice. Oncotype DX thus represents an alternative to Prosigna for node negative patients, but Oncotype DX was not compared with Prosigna in the submitted documentation or in this STA. No gene-profiling test is currently recommended for node positive patients in Norway. Oncotype DX may thus fulfill an unmet need for postmenopausal patients with node positive disease. Results from Oncotype DX may also be combined with traditional clinical parameters such as tumor grade, tumor size, proliferation status, and lymph node status, but it is not clear how the different parameters should be weighted in possible combinations.
The cost-effectiveness and budget impact analyses were conducted by integrating various sources of evidence and assumptions, which may have contributed to overall uncertainty in the model. The most important sources of uncertainty were regarding the distribution of RS, and that the different studies used different thresholds for which patients to offer adjuvant chemotherapy.
Conclusion
Oncotype DX predicted chemotherapy benefit in patients with ER+ HER- early-stage breast cancer who were node negative (regardless of menopausal status) or postmenopausal and node positive (1-3 lymph nodes). In these groups, patients with low or intermediate RS (0-25) did not show chemotherapy benefit and could omit chemotherapy to reduce side-effects, whereas patients with high RS (>25) showed chemotherapy benefit and should be offered chemotherapy to reduce the risk of recurrence. The distribution of RS in breast cancer populations suggested that chemotherapy use can be substantially reduced. Decision-impact studies demonstrated that Oncotype DX can reduce chemotherapy assignment in clinical practice, but also illustrated that treatment decisions were not based entirely on RS.
Oncotype DX seems to be more effective and less costly compared to no gene-profiling test. Sensitivity analyses confirmed that Oncotype DX is probably cost-effective, also at low thresholds of willingness-to-pay. As the two tests were not compared, it remains unclear whether Oncotype DX is more cost-effective than Prosigna for node negative patients in Norway.
The budget impact analysis for node negative patients indicate incurred net costs in the five years after implementation, but this analysis is of limited relevance since Oncotype is compared to no gene-profiling test, rather than Prosigna. Implementation of Oncotype DX for postmenopausal lymph node positive patients seems to be cost saving the first five years.
