Hopp til innhold

Selected items added to basket

Go to basket

Health technology assessment

Prosigna Gene Signature to Assess Expected Benefit from Chemotherapy in Breast Cancer. Assessment of manufacturer’s submission

  • Year: 2019
  • By: Norwegian Institute of Public Health
  • Authors Fagerlund BC, Chudasama KK, Brurberg KG, Robberstad B, Rose C, Juvet LK, Fretheim A.
  • ISBN (digital): 978-82-8406-020-0
Forside_Prosigna ENG.jpg

In this health technology assessment, we have considered a molecular profiling panel, Prosigna, which is meant to improve the assessment of recurrence risk among women who have undergone surgical treatment for breast cancer.

Downloadable as PDF. In English. Norwegian summary.

Have you found an error?

Order

Download:

Key message

Background

Breast cancer can be treated with chemotherapy, hormone therapy and radiation, or a combination of these to prevent the spread of cancer cells, after surgical removal of the tumor. When assessing whether a patient should be offered chemotherapy, information about prognosis is important. Patients at high risk of recurrence should be offered chemotherapy, while patients at low risk are not very likely to gain from such treatment, in which case side effects outweigh the benefits. The assessment of risk of recurrence is based on clinical findings, e.g. tumor size, lymph node involvement, and expression of certain receptors on the cancer cells.

In this health technology assessment, we have considered a molecular profiling panel, Prosigna, which is meant to improve the assessment of recurrence risk among women who have undergone surgical treatment for breast cancer.

Our assessment is based on documentation submitted by the manufacturer of Prosigna, Nanostring.

Objective

The objective was to investigate the prognostic accuracy, clinical effectiveness, and cost effectiveness of Prosigna in patients diagnosed with breast cancer.

In Norway, the group considered as potentially eligible for the test is patients with breast cancer who had their tumor removed, are node-negative and where the tumor is classified as hormone receptor-positive/human epidermal growth factor receptor 2-negative (HR+/HER2-). 

Methods

Prognostic accuracy and clinical effectiveness

To validate the submitted evidence we extracted data from the key publications and critically appraised the risk of bias in the findings.

Health economics

We assessed cost‐effectiveness estimates of Prosigna compared with current practice for HR+/ HER2-, node negative patients, provided by the submitter, and similarly for a defined subgroup of patients at higher risk of recurrence (“Luminal B like pT1c-pT2 pN0”).  The estimates were based on a hybrid model with a decision tree combined with a Markov model. Cost-effectiveness estimates for Prosigna versus current practice were modelled over a 50-year time horizon, for patients aged 58. We performed separate analyses using the submitted model, adjusting some of the input variables based on revised assumptions. Also, we performed alternative scenario analyses for differences in chemotherapy use with and without use of the Prosigna test, based on various other data sources.

Results

Prognostic accuracy

There is convincing evidence of a correlation between the observed risk of recurrence and the risk stratification score generated by the Prosigna test. For patients classified as low risk, the ten-year risk of recurrence is around 4%. For the intermediate risk group, the risk is around 10%, and for the high-risk group around 21%. We also expressed the performance of the test in terms of prognostic sensitivity and specificity. When we merged the intermediate risk group with the low risk group (intermediate test constitutes a “negative” test), we estimated the test’s sensitivity to 52%, and its specificity to 77%. If the intermediate group was merged with the high-risk group (intermediate risk constitutes a “positive” test), the sensitivity and specificity were estimated to be 83% and 42%, respectively.

The estimates presented above reflect the test’s performance when used as a standalone tool. In practice, it can be anticipated that the test will be used as a supplement to the current risk stratification approaches.

Clinical effectiveness

We did not identify comparative studies where patients were allocated to risk stratification with or without the Prosigna and followed over time. Without such comparative studies, it is difficult to estimate the clinical utility of Prosigna, i.e. Prosigna’s impact on the use of chemotherapy and patient outcomes such as disease-free survival and side effects from chemotherapy. Studies exploring the prognostic value of adding Prosigna to other prognostic variables in multivariate regression models suggest that Prosigna adds prognostic information that may be useful when deciding about further use of chemotherapy. However, these data are sparse, and it remains unclear to what extent Prosigna will contribute to fewer recurrences or a reduction in the needless use of chemotherapy than current practice.

Health economics

The incremental cost-effectiveness ratio (ICER) based on the revised economic model for HR+/ HER2-, node negative patients, is calculated to NOK 897,923 per QALY gain in our base-case analysis. The estimate is based on questionable assumptions and is highly sensitive to changes in the chemotherapy use parameter. We estimated the total added costs of implementing Prosigna for this group in Norway, to about NOK .5 million in year five.

The calculated incremental cost-effectiveness ratio (ICER) based on the revised economic model for the subgroup of “Luminal B like pT1c-pT2 pN0”-patients (38% of the “all node negative”-population) would be NOK 98,188 per QALY gain. Implementing Prosigna for this subgroup in Norway, would lead to a total cost saving of NOK 9.9 million in year five.

Discussion

Clinical efficacy and safety

Several studies have assessed the extent to which the test is able to categorise patients into groups with a low, intermediate or high risk of recurrence. However, the utility of this information for clinical decision making is uncertain. Uncertainties are mainly caused by lack of data regarding the accuracy of procedures that are currently used when selecting patients for chemotherapy, and uncertainties regarding the emphasis clinicians and patients will put on Prosigna when deciding for or against chemotherapy. The relatively low sensitivity of the Prosigna test means that it yields a considerable number of false negative classifications, which entails a risk that patients who could benefit from chemotherapy are not offered the treatment.

Evidence from multivariate regression analyses indicate that Prosigna contributes information of prognostic value beyond tests and assessment tools in current use. The manufacturer of Prosigna has not based any of the analyses in the submission on these data, and we did not see how these results could be used to estimate an expected health gain from introducing Prosigna testing into clinical practice.

Health economics

Regarding the model input, empirical data on chemotherapy use is lacking. In the submitted model the proportion of chemotherapy use following Prosigna testing was based on the opinions of 11 British oncologists. The proportion of chemotherapy use in current practice (no test) was assumed to be the same across the different risk of recurrence-groups – a dubious assumption.

We used data derived from a Norwegian study and recommendations in Norwegian clinical practice guidelines for breast cancer management, which we believe yield more trustworthy estimates than those in the submitted analyses.

Further, there is controversy regarding the utility value of “the risk profiling knowledge to patients”, which was assumed by the submitter. We do not consider preferences for knowing the test result “health-related”, and we therefore consider this parameter irrelevant in this case.

The economic model submitted by NanoString did not incorporate sensitivity and specificity, and we are uncertain what the consequences of this are for the cost-effectiveness-estimates. If we were to prepare a health economic model for a prognostic test such as Prosigna, we would probably have opted for a different approach, and included the test’s prognostic sensitivity and specificity into the model.

Conclusion

There is probably a statistical association between Prosigna's risk prediction and the observed risk of distant recurrence after breast cancer. However, it is uncertain to what extent Prosigna contributes prognostic information that translates into better clinical results in terms of lower recurrence rates and reduced chemotherapy use

Conclusions about the cost-effectiveness of Prosigna cannot be made as we do not have reliable data on chemotherapy use and clinical outcomes for patients who have or have not undergone Prosigna testing.