6 Considerations

6 Considerations

6.1 The Diagnostics Advisory Committee discussed the focus of the evaluation and the evidence available for the 4 tests. It noted that gene expression profiling and immunohistochemistry tests other than those included in this evaluation are being developed. The Committee also noted that at present, the level and quality of the available evidence varies for the 4 tests. In particular, evidence on the tests' ability to guide clinical decisions on the use of chemotherapy in England and to predict response to chemotherapy in women with early breast cancer was limited. The External Assessment Group's economic model for women with oestrogen receptor positive (ER+), lymph node negative (LN−) or human epidermal growth factor receptor 2 negative (HER2−) early breast cancer was used by the Committee when considering the likely cost effectiveness of the 4 tests. The Committee considered that the most appropriate use of these tests is in women for whom the decision to offer chemotherapy is uncertain, that is, women at intermediate risk of distant recurrence. It therefore considered that the subgroup analyses of women with a Nottingham Prognostic Index (NPI) score above 3.4 were the most relevant, based on the likelihood that there would not be many women with an NPI score of above 5.4 within the target population.

6.2 The Committee acknowledged the emotional and psychological strain for patients with breast cancer when considering therapy, in particular, chemotherapy and its associated adverse events. The Committee noted that this is likely to be significant in patients for whom the decision about whether or not to have chemotherapy is difficult after receiving the results of current tools used in the NHS (especially patients deemed to be at intermediate risk). The Committee also noted that tools used by the NHS to assess the suitability of patients with breast cancer for adjuvant chemotherapy vary across England. The Committee concluded that any tests that can help to alleviate emotional and psychological strain and promote consistency of practice within the NHS are likely to be appreciated by patients and clinicians alike.

6.3 The Committee discussed the generalisability of the data to men. The Committee acknowledged that breast cancer is not only observed in women and that men make up a small proportion of patients with breast cancer. The Committee noted that all the clinical and economic evidence had been based on trials with women; however, experts on the Committee stated that even though there are some subtle gender-specific differences in the pathobiology of breast cancer, the general subtypes are identical in men and women. Therefore, in clinical practice men would be treated in the same way as women. The Committee therefore concluded that the recommendations in this guidance should also apply to men.

6.4 The Committee discussed the evidence base for Oncotype DX and concluded that, in general, it was the most developed of the 4 tests in the evaluation. The Committee discussed the analytical validity of Oncotype DX. The Committee noted that no new evidence was identified in the External Assessment Group review, but that evidence was identified in the previous systematic review (Marchionni et al. 2008) that showed reasonable within-laboratory replicability. The Committee also noted that the test is processed centrally by the manufacturer in the USA and the laboratory is CLIA (Clinical Laboratory Improvement Amendments) certified. Given the above, the Committee was satisfied with the analytical validity of the test. The Committee discussed the prognostic ability (clinical validity) of Oncotype DX. Experts on the Committee pointed out and the Committee agreed that the prognostic ability (the ability to predict the risk of distant recurrence) of Oncotype DX had been well validated. The Committee also considered a study by Sgroi et al. reported in abstract form at the San Antonio Breast Cancer Symposium 2012, which assessed the prognostic value of Oncotype DX (and IHC4) over and above standard clinical variables. The Committee noted that the abstract shows that Oncotype DX does not provide prognostic information for late distant metastasis. The Committee considered that these new data raise potential uncertainty around the long-term benefits of Oncotype DX, but judged that the relatively extensive evidence base supporting the prognostic ability of the test to be satisfactory at this time. The Committee therefore concluded that the prognostic ability of Oncotype DX was supported by robust evidence in the early breast cancer population.

6.5 The Committee then discussed the clinical utility of Oncotype DX. It heard from the External Assessment Group that a key aspect of clinical utility is the ability of a test to accurately predict those patients who will benefit most from chemotherapy. The Committee therefore considered whether gains from chemotherapy could differ between patients with different prognoses (that is, patients in different risk groups). Experts on the Committee pointed to data from recent meta-analyses that showed proportional gains from chemotherapy were generally constant across clinical parameters such as tumour diameter and ER status (used to help determine a patient's prognosis). However, these constant proportional gains meant that those with a good prognosis would receive less absolute benefit from chemotherapy than those with a poor prognosis. Furthermore, the possibility that chemotherapy might be more effective both proportionally and absolutely in patients identified by Oncotype DX, given that the test provides information about the biological features of the tumour, was discussed. The possibility that tumours with the genomic characteristics identified by Oncotype DX might be more susceptible to chemotherapy was also explored. The evidence on the predicted benefit of chemotherapy (reduction in the risk of distant recurrence) for women receiving chemotherapy in addition to endocrine therapy compared with endocrine therapy alone was discussed. The Committee heard that data were available that suggest that Oncotype DX can predict the relative benefit of chemotherapy and that the effectiveness of chemotherapy varies depending on the classification of patients by the Oncotype DX test in LN− patients (Paik et al. 2006). These data indicated that women in lower risk groups benefit proportionally less from chemotherapy than those in higher risk groups (see section 5.14). The Committee considered that the Paik study was limited by its design, the sample sizes of individual risk groups, the use of some results from the training dataset (tamoxifen-treated patients of the NSABP B‑20 trial) in the study dataset, the applicability of the study population (a younger population that includes patients with HER2+ breast cancer) to the population considered in this guidance, and the fact that the treatments (endocrine therapy and chemotherapy) used are different to those currently used in the NHS. In addition, the Committee considered that the relative benefit from chemotherapy by risk group was unclear. The Committee concluded that the evidence implying a predicted differential relative benefit of chemotherapy according to Oncotype DX risk group in LN− patients (Paik et al. 2006) was not robust. The Committee also reviewed evidence implying a predicted differential relative benefit of chemotherapy according to Oncotype DX risk group in LN+ patients (Albain et al. 2010) and data from the neoadjuvant setting. The Committee concluded that these data were not robust enough to support the test's ability to predict the benefit of chemotherapy. In the absence of robust data the Committee concluded that equal benefit of chemotherapy should be assumed across all Oncotype DX risk groups. Therefore, although the Committee considered that adequate evidence supported the prognostic ability of Oncotype DX (that is, its ability to predict the risk of distant recurrence, see section 6.4), it concluded that it was not confident in the ability of Oncotype DX to predict benefit from chemotherapy.

6.6 The Committee discussed the cost effectiveness of Oncotype DX based on the original price proposed by the manufacturer (list price). The Committee considered that the incremental cost-effectiveness ratios (ICERs) from the base-case analysis of Oncotype DX were not the most appropriate for decision-making purposes because of the assumption of a predicted differential relative benefit of chemotherapy according to Oncotype DX risk group. The Committee discussed the ICERs presented in the sensitivity analysis that assumed equal benefit of chemotherapy across all Oncotype DX risk groups (at a level of either 30% or 40% relative risk reduction from chemotherapy). The Committee noted the ICERs for Oncotype DX (compared with current practice) when offered to all women with ER+, LN−, HER2− early breast cancer were £91,300 (30% relative risk reduction from chemotherapy) per quality-adjusted life year (QALY) gained and £64,900 (40% relative risk reduction from chemotherapy) per QALY gained. The Committee considered the ICERs to be too high to recommend Oncotype DX for use in the NHS for all women with ER+, LN−, HER2− early breast cancer. The Committee considered that the overall benefit of chemotherapy was likely to be closer to 27% relative risk reduction from chemotherapy across all Oncotype DX risk groups (EBCTCG overviews 2005, 2011). The Committee noted that there are potential differences in the population included in the EBCTCG review compared with the population in the economic analysis and that the outcome measures differed. Without data specific to the population under consideration, the Committee considered the EBCTCG figure to be the most appropriate for use at this time. The Committee concluded that the most plausible ICER, based on the evidence presented, was likely to exceed £91,300 for all women with ER+, LN−, HER2- early breast cancer. Therefore, based on the original proposed price (list price), Oncotype DX would not be a cost-effective use of NHS resources in this group.

6.7 The Committee then considered a proposal submitted by the manufacturer of Oncotype DX. The proposal makes Oncotype DX available to the NHS at a revised price for those people assessed as being at intermediate risk. The proposed price is commercial in confidence. NICE advised the Committee, and the Committee agreed, that the access proposal appeared workable and efficient, and did not appear to constitute an excessive administrative burden on the NHS. The Committee went on to discuss the impact of the proposal on the cost effectiveness of Oncotype DX in people assessed as being at intermediate risk when it was assumed that Oncotype DX was able to predict a patient's prognosis but not the benefit of chemotherapy (relative risk reduction of distant recurrence from chemotherapy). The Committee accepted an analysis performed by the External Assessment Group, which showed that the ICER for Oncotype DX (compared with current practice) in this group of patients was £22,600 per QALY gained, assuming prognostic benefits of the test but no predictive effect. The Committee also noted the ICER could be significantly lower if Oncotype DX was shown to predict the benefit of chemotherapy by robust evidence from future research. The Committee noted that an NPI score above 3.4 was used in the analysis as a mechanism for identifying patients at intermediate risk, but also noted that other methods for determining the risk group were available and in use in the NHS. The Committee believed that the subgroup analysis of people with an NPI score above 3.4 was likely to be a reasonable approximation for people at intermediate risk generally. Therefore, given the strength of the evidence on the prognostic ability of the test (and evidence of analytical validity), the Committee concluded that Oncotype DX for use in people at intermediate risk of distant recurrence, when the decision to prescribe chemotherapy remains unclear, would represent a cost-effective use of NHS resources if acquired at the confidential revised price offered by the manufacturer.

6.8 The Committee discussed the need for further robust evidence to demonstrate the ability of Oncotype DX to identify patients who will benefit most from chemotherapy (see section 6.5). The Committee considered that further information on the clinical utility of the test is warranted. This should comprise the development of robust evidence on the impact of Oncotype DX on clinical decision-making in England. The Committee noted that the Oncotype DX score may be combined with existing clinicopathological variables used informally by physicians at the local level, or more formally using a pre-specified algorithm. These 2 approaches should be kept in mind for any future research on the impact of Oncotype DX on clinical decision-making in England. The Committee noted that a decision-impact study in Bristol is near completion. Research should also address the ability of the test to predict the benefit of chemotherapy. The Committee noted there is an ongoing prospective trial (TAILORx) that will provide further information on the benefit of chemotherapy in women classified as intermediate risk by Oncotype DX. As the patient population included in TAILORx is from North America, the Committee encouraged the collection of data on Oncotype DX when used in the NHS in England (see section 7). The Committee was mindful that the extra value of the tests when used in addition to current clinical practice has been shown by the model constructed by the External Assessment Group, and that the use of Oncotype DX in the NHS in England represents an opportunity to collect further data on this. The Committee concluded that multicentre audit should be a priority for further investigation.

6.9 The Committee discussed the evidence available on the analytical validity of IHC4. It noted the test was at a comparatively early stage of development. In particular, it was noted that although there are data on the reliability and reproducibility of the measurement of ER, progesterone receptor (PR) and HER2 markers, data were lacking on the reliability and reproducibility of the Ki‑67 marker measurement. The Committee heard that ER, PR and HER2 have an established UK National External Quality Assessment Scheme (NEQAS), and that a study was published recently on the reproducibility of Ki‑67 and a UK NEQAS was being investigated for the marker. The Committee noted however that quantitative assessments of ER, PR and Ki‑67 (not routinely reported in the NHS) should be appropriately considered in the NEQAS if not already done so. The Committee considered that data are needed on the reproducibility and reliability (analytical validity) of the complete IHC4 test (an algorithm combining 4 markers and classical clinical and pathological variables). This is particularly important as the test is designed for local processing in NHS laboratories. An additional study on quality assurance was also considered by the Committee. This was a small preliminary study that did not materially change the results of the External Assessment Group analysis. The Committee concluded that the lack of data on analytical reliability meant that it was not possible to make a recommendation for general use of the IHC4 test at this time.

6.10 The Committee then discussed the clinical validity and clinical utility of IHC4. It noted that only 1 study was available on the clinical validity of the test. The Committee discussed the separate cohort of 786 patients used for external validation of the test in this study and concluded that it was not fully representative of the population of interest in this assessment because approximately 50% of patients did not receive 5 years of endocrine therapy. The Committee also noted that the External Assessment Group review did not identify any data on the clinical utility of IHC4. An additional study on how the test classifies patients by risk group compared with the NPI and Adjuvant! Online was also considered by the Committee. Although encouraging, this was a small preliminary study that did not provide an indication of how management decisions would actually change and did not materially change the results of the External Assessment Group analysis. The Committee also noted the recent availability of further data on IHC4 in the large TEAM study. The Committee considered that the general uncertainty in the clinical effectiveness evidence for IHC4 limited the validity of the economic analysis. It concluded that robust data on how the test might be used in the NHS in England (a continuous risk score, defined risk groups or both) and the impact of the test on clinical decision-making are needed. The Committee also indicated that data on the benefit of chemotherapy according to IHC4 score (or defined risk groups) would be useful. Although IHC4 was found to dominate current practice in the base-case economic analysis when offered to all women with ER+, LN−, HER2−early breast cancer and in a subgroup of women with an NPI score above 3.4, in addition to most of the sensitivity analyses, the Committee concluded that the uncertainty in the estimates of the clinical effectiveness of the test was too great to recommend adoption at this time. The Committee considered that further evidence was needed before the test could be adopted for general use by the NHS. Given the estimated low cost of the test and the modelling results that showed it has the potential to dominate current practice, the Committee considered it prudent to recommend the use of IHC4 for research in the NHS to collect information on the analytical validity, and hence, clinical validity and clinical utility of the test (see section 7).

6.11 The Committee discussed the clinical evidence and the uncertainty in the estimates of the cost effectiveness of Mammostrat (because of uncertainty in the clinical evidence underpinning the economic analysis). The Committee noted there were limited published data on the analytical validity of the test. It went on to discuss the clinical validity of the test, and considered the results of the economic analysis to be limited because the risk reclassification data (provided in confidence) derived from a small subset of women included in the study by Ring et al. (2006; US cohort) demonstrated some inconsistencies and were not sufficiently robust in demonstrating the ability of the test to predict which women were at low, intermediate or high risk in the subgroup of women with an NPI score above 3.4. The Committee noted the recent availability of data on the clinical validity of Mammostrat in the TEAM study. The Committee considered that this study provides additional supportive data for a large UK population on the prognostic ability of the test and expands the evidence base to patients treated with aromatase inhibitors, rather than tamoxifen. Although further supportive data are available on the clinical validity of the test, the Committee considered that the economic analysis was also limited because there was uncertainty about the clinical utility of the test; in particular no evidence exists on how the test would affect clinical decision-making in England, and because of the discordant results (Ross et al. 2008) on the benefit of chemotherapy (only the low- and high-risk groups benefitted from chemotherapy but not the intermediate-risk group). Overall, although a limited number of studies have been conducted, the Committee was encouraged by the large sample sizes of the studies showing that Mammostrat can act as an independent prognostic tool. In particular, the Committee noted that a significant portion of the prognostic evidence was generated using UK-based patients. However, the Committee felt that to fully understand the benefits of the test, further data are needed to demonstrate how the test reclassifies people's risk when compared with current practice in England, and to demonstrate the impact of Mammostrat on clinical decision-making in England. The Committee considered that the uncertainty in the clinical-effectiveness evidence for Mammostrat limited the validity of the economic analysis. Therefore, given the uncertainty in the clinical effectiveness of the test (in particular, the analytical validity and clinical utility), the Committee was unable to recommend the adoption of Mammostrat for general use in the NHS at this time and recommended the test for research use only. The Committee heard that there is an extensive ongoing research programme for this relatively new test.

6.12 The Committee discussed the clinical evidence and the uncertainty in the estimates of cost effectiveness of MammaPrint. The Committee noted that although the MammaPrint test was created using samples from an untreated breast cancer population, in particular with samples from patients who had not received endocrine therapy, data were available on the use of the test in patients treated with adjuvant endocrine therapy (for example, Kok et al. 2010). The Committee was not aware of evidence on the use of MammaPrint in UK clinical practice. The Committee noted that no new evidence was identified in the External Assessment Group's review on the analytical validity of the test, but that evidence had been identified in the previous systematic review by Marchionni et al. (2008). The Committee went on to consider the different sample types used by the test. The Committee was aware that the use of MammaPrint on formalin-fixed paraffin-embedded samples has been CE marked and that the manufacturer had submitted data to the Food and Drug Administration to demonstrate that the performance of MammaPrint in formalin-fixed paraffin-embedded samples is equivalent to that of fresh samples. The Committee discussed the clinical validity of the test and agreed with the External Assessment group that such evidence, although developing, is based on cohort studies of small sample sizes that have been conducted outside of England in a heterogeneous population of predominantly younger pre-menopausal women (younger women are more likely to be classified as having a poor prognosis using MammaPrint, which may overestimate the benefit of the test in the early breast cancer population as a whole). The Committee discussed the clinical utility of the test and noted that the risk reclassification data and the proportion of patients receiving chemotherapy were taken from studies of Dutch patients that included predominantly pre-menopausal women. Furthermore, the Committee agreed with the External Assessment Group that the impact of the test on decision-making in England had not been demonstrated and that the Knauer et al. study (2010) had considerable methodological limitations (see section 5.12), and therefore the Committee considered that the clinical utility of MammaPrint had not been robustly demonstrated. The Committee considered that the uncertainty in the clinical-effectiveness evidence for MammaPrint limited the validity of the economic analysis.

6.13 The Committee also considered additional evidence on MammaPrint forwarded by the manufacturer, including the RASTER study (an updated analysis by Drukker et al. 2013), the IMPAKT 2012 working group statement and a summary of cost-effectiveness results based on NPI scores from 2 patient series data. However, the Committee agreed with the External Assessment Group, which concluded that this evidence did not materially change the results of the analysis. The Committee noted that the RASTER study provided prospective data on the additional prognostic value of MammaPrint when compared with Adjuvant! Online, in the form of an observational study of over 400 patients. However, the Committee considered that this study did not substantially reduce the uncertainty in clinical effectiveness because RASTER, similarly to other studies of MammaPrint, was conducted in the Netherlands in a younger population than that seen in England, with most patients younger than 55 years. The Committee also noted that treatment decisions were based on a range of factors, including the Dutch Institute for Healthcare Improvement (CBO) guidelines that are not used in England. In addition, the RASTER study included some patients with ER− (20%) or HER2+ (11% positive and 5% unknown) breast cancer. Using MammaPrint, these patients would be very likely to be categorised as having a poor prognosis and receive chemotherapy, which may lead to an overestimation of the benefit of the test in the population considered in this evaluation (ER+, LN− and HER2−). The Committee also considered the consensus statement by the IMPAKT 2012 working group and noted that it found the available evidence on the analytical validity and clinical validity of MammaPrint to be convincing. The Committee noted that the consensus statement did not summarise any new evidence not already included in the External Assessment Group's report. The Committee considered that the data on the clinical validity of the test were not generalisable to the population considered in this evaluation for the reasons already stated (that is, the data were from cohort studies of small sample sizes conducted outside England in a heterogeneous population of predominantly younger pre-menopausal women). The Committee agreed with the External Assessment Group that the cost-effectiveness results based on NPI scores from 2 patient series lacked detailed descriptions of methodology, and so did not allow a clear assessment of the quality of the evaluation. Therefore, the Committee considered that the uncertainty in the clinical effectiveness data remained, and limited the validity of any economic analysis despite the additional evidence.

6.14 The Committee concluded that the uncertainty in the clinical effectiveness, in particular the clinical validity and clinical utility of the test, was too high to recommend the adoption of MammaPrint for general use in the NHS at this time and recommended the test for research only. The Committee noted that there is an ongoing prospective clinical trial (MINDACT) on the value of MammaPrint in predicting which patients would benefit from chemotherapy and that results from this trial may help to reduce the uncertainty about effectiveness.

6.15 The Committee expressed general concern over the lack of information on the impact of the use of gene expression profiling and expanded immunohistochemistry tests on clinical decision-making in England. It noted that some limited data on clinical decision-making in England were available for Oncotype DX, which were helpful in informing the assessment. The Committee requested further data on the ability of the tests to impact clinical decision-making. The applicable data forwarded by the manufacturers came from studies that were conducted outside England. The Committee agreed with the External Assessment Group that a lower baseline level of chemotherapy prescribing in England than in the USA or many other European countries increases the uncertainty in the generalisability of studies conducted outside England.

6.16 A potential equality issue was raised by the Committee, which was concerned about the lack of evidence on the use of the tests in women older than 75 years. The Committee accepted the evidence had been limited to women younger than 75 years; however, the recommendations in section 1 do not restrict access to the tests based on age of the patient. The Committee also discussed potential equality issues concerning the use of the new tests in men. The Committee heard that given the relatively low number of men with breast cancer when compared with women, evidence on the performance of these tests in men was less likely to be generated. Experts on the Committee pointed out that breast cancer in men shared many characteristics with that seen in women and that both groups were treated similarly in clinical practice (see section 6.3). Therefore, the Committee felt it appropriate that the recommendations apply to both men and women.

  • National Institute for Health and Care Excellence (NICE)