EGFR‑TK mutation testing in adults with locally advanced or metastatic non-small-cell lung cancer

The committee considered the technical performance of the different tests. It heard from clinical specialists on the committee that, in their experience, the different tests generally have a similar level of accuracy in detecting the mutations that they are designed to detect. The committee also noted the statement from the UK National External Quality Assessment Service (NEQAS) that errors seen in the EGFR quality assurance scheme are not always method related, and that variations in how tests are processed and implemented can lead to variations in the failure rates (see section 5.11). The committee also considered the failure rates reported for the different tests in the web-based survey. The committee considered that, although the survey was limited by its small sample size, it seemed to suggest that failure rates are generally not test-dependent. Furthermore, the committee heard that it is standard practice for the quality of tissue samples to be initially assessed by a pathologist. Therefore, the decision made by the pathologist about whether to send a sample for EGFR‑TK mutation testing could impact on the number of patients with an unknown EGFR‑TK mutation status. The committee concluded that the technical performance of the tests is not solely influenced by test accuracy, and that processing of samples and testing practices is likely to influence technical performance. It further concluded that, in UK practice, the technical performance of the tests under assessment is likely to be very similar.

The committee discussed the lack of a gold standard test for assessing test accuracy, the difficulties relating to the different mutation coverage of the various tests, and the uncertainty about the clinical significance of some mutations. The committee acknowledged that the approach taken by the external assessment group, in which accuracy for predicting response to treatment was calculated using objective response and disease control as reference standards, was a valid approach in this situation. It was advised by clinical specialists on the committee, however, that deriving accuracy from response to treatment with an EGFR‑TK inhibitor is problematic. For instance, the definition of false positives was 'patients identified as having tumours with an EGFR‑TK mutation that do not respond to treatment with an EGFR‑TK inhibitor'. It noted that there may be other reasons why a tumour does not respond to treatment, such as concomitant medications, patient characteristics and other clinical factors.

The committee considered the accuracy of EGFR‑TK mutation tests for predicting response to treatment. It noted that the external assessment group had only been able to obtain sensitivity and specificity estimates for therascreen EGFR PCR Kit and Sanger sequencing of exons 18 to 21 and exons 19 to 21. The committee also noted that, when accuracy estimates were available, the accuracy of different tests was calculated from different studies with different patient populations and different ways of classifying resistance mutations. The committee heard from the external assessment group that, ideally, sensitivity and specificity values should be generated for all tests from a single set of samples, therefore limiting the influence of sampling and population differences on the accuracy estimates. The committee acknowledged that this assessment did not present such a scenario and that it was plausible that differences in patient populations could have impacted on relative accuracy estimates for individual tests. The committee therefore concluded that the relative predictive accuracy for the different tests could not be reliably established.

The committee considered the effect of tissue quality on the accuracy of EGFR‑TK mutation testing. It heard from clinical specialists on the committee that the quality of tissue samples available for testing often varies, and this may impact on both the test failure rates and test accuracy. The committee noted that, in addition to obtaining good-quality tumour samples, it is important to use a sensitive test to enable detection of EGFR‑TK mutations, especially to ensure correct results in lower-quality tumour samples. It discussed the consequences of assigning the wrong EGFR‑TK mutation status to a patient and noted that both the IPASS and the Signal-FIRST trials (see section 5.21) had demonstrated that progression-free survival was shorter for patients receiving an EGFR‑TK inhibitor than for patients receiving standard chemotherapy in the EGFR‑TK mutation-negative subgroup. For this reason, the committee concluded that it is important to ensure high accuracy of testing, particularly to minimise the chances of incorrect treatment.

The committee then discussed the generalisability of the clinical evidence to UK clinical practice and the UK patient population. It noted that 4 out of the 5 randomised controlled trials identified by the external assessment group were conducted in East Asia (see sections 5.18 and 5.19). The committee acknowledged that the patients included in the trials had characteristics different from patients usually seen in UK practice, most notably that the studies included a high proportion of patients who had never smoked and a high proportion of patients of East Asian origin. The committee noted that all evidence for the therascreen EGFR RGQ PCR Kit came from the IPASS trial, which looked almost exclusively at patients from East Asia. It also noted that, although most of the evidence came from patients with adenocarcinoma, patients with squamous cell carcinoma would also be tested for EGFR‑TK mutations and these patients may have different clinical characteristics. The committee concluded that, although there were some substantial differences between the trial populations and the population of patients presenting with advanced NSCLC in the UK, the effect on test accuracy was likely to be minimal and therefore the trial evidence could be used to support the effectiveness of testing in patients with adenocarcinoma and in patients with squamous cell carcinoma in a UK setting.

The committee considered the value of tests that identify rare EGFR‑TK mutations. It acknowledged that screening tests are designed to detect more mutations than the targeted tests. In addition, the targeted tests (therascreen EGFR PCR Kit and the cobas EGFR Mutation Test) are designed to detect different sets of mutations, which vary in the number of rare forms of mutations included. The committee heard from clinical specialists on the committee that the clinical significance of rare mutations is generally unknown, and that treatment decisions for patients with a rare EGFR‑TK mutation would be made by the oncologist based on the availability of evidence, such as case studies. The committee acknowledged that evidence on the clinical effect of rare mutations is being generated. However, it concluded that currently there is little additional value of tests designed to detect rare mutations, except for the purpose of collecting clinical outcome data for research.

The committee discussed whether there are any benefits of using CE‑marked tests over laboratory-developed tests for detecting EGFR‑TK mutations. It heard that the CE‑marked EGFR‑TK mutation tests (therascreen EGFR RGQ PCR Kit and cobas EGFR Mutation Test) and the simpler laboratory-developed tests such as polymerase chain reaction (PCR) may be easier to implement than tests based on Sanger sequencing for laboratories with little molecular diagnostics experience (for example, pathology laboratories). The committee therefore concluded that, although there was no distinguishable difference in the technical performance of the tests, the ease of use of the CE‑marked tests may be an advantage in some clinical settings, particularly when limited molecular diagnostics technical support is available.

The committee considered the turnaround time of EGFR‑TK mutation testing. It noted that turnaround time was assumed not to be test-dependent and was therefore not included in the economic modelling. The committee heard from clinical specialists on the committee that turnaround time was impacted by factors such as transporting samples between different locations for testing and the set-up of the laboratory. It also heard from patient experts that waiting for test results causes additional anxiety to patients and that the rapid turnaround of test results is a priority for patients. The committee acknowledged that, although the survey conducted by the external assessment group showed, on average, that similar turnaround times were achieved for all tests in the UK laboratories surveyed, the only test achieving a turnaround time shorter than 3 days was the therascreen EGFR PCR Kit. However, the committee concluded that, although it is possible that the CE‑marked tests could achieve quicker turnaround times, the frequency of batch testing would have a considerable impact on turnaround time in practice, and therefore it is likely that standard turnaround times could be met, irrespective of which test method is used.

The committee considered next-generation sequencing and noted that research is currently being done on this method to look at panels of lung cancer genes. It noted that current turnaround time and cost are a hindrance to implementation, but that these practicalities are likely to be resolved in the future. The committee concluded that next-generation sequencing is likely to be an important method for identifying EGFR‑TK mutations in the future.

The committee noted that the price a laboratory charged for an EGFR‑TK mutation test was used in the cost-effectiveness analyses and that this price is not necessarily a true reflection of the actual cost to a laboratory. The committee noted that the mean test costs reported in the survey of laboratories in England and Wales ranged from £130 to £188 (see section 5.27). It heard from clinical specialists on the committee that the true cost to a laboratory may vary depending on their individual set-up, and that costs can change over time as experience and throughput changes. The committee accepted that the reference case states that costs to the NHS should be used, and therefore that the approach taken by the external assessment group in their cost-effectiveness modelling was appropriate. The committee concluded that the true costs of the tests are likely to be very similar for all the tests included in this assessment and that they are appropriately incorporated in the cost-effectiveness models.

The committee noted that the overall survival estimates used in the 'comparative effectiveness' and the 'linked evidence' cost-effectiveness analyses came from the IPASS trial for the therascreen EGFR PCR Kit and from the First-SIGNAL trial for Sanger sequencing. The committee heard from the external assessment group that the reason for this was the need to use the same assumptions as in NICE technology appraisal guidance on gefitinib for the first-line treatment of locally advanced or metastatic non-small-cell lung cancer. It also heard from the external assessment group that, if the IPASS survival estimates were used for both tests, Sanger sequencing was more costly and more effective than the therascreen EGFR PCR Kit. However, if the survival estimates from the First-SIGNAL trial were used for both tests, the therascreen EGFR PCR Kit became more costly and more effective than Sanger sequencing. The committee noted that that 'comparative effectiveness' and the 'linked evidence' cost-effectiveness models appeared sensitive to the difference in quality-adjusted life year (QALY) gains from the 2 trials, and that these results could be because of relatively small differences in QALYs.

The committee considered the assumptions used in the cost-effectiveness analyses. It noted that, in the 'comparative effectiveness' and the 'linked evidence' analyses, a key assumption was that the difference in comparative effectiveness between the studies was solely because of the use of different tests. However, the committee acknowledged that the differences in comparative effectiveness between the tests may be caused by a variety of factors, such as differences in the patient populations. The committee therefore concluded that the assumption on comparative effectiveness used in these models is unlikely to hold true.

The committee considered the face validity of the 'comparative effectiveness' and the 'linked evidence' analyses. It noted that, although the external assessment group had made a good attempt to model the cost effectiveness of EGFR‑TK mutation testing, the analyses were severely hampered by lack of evidence and therefore the greater level of assumptions needed. The committee was especially concerned about the uncertainties in input parameters, such as cost and overall survival (see sections 6.12 and 6.13). The committee concluded that, given these problems, the results of the cost-effectiveness analyses were not robust.

The committee considered the validity of the results of the 'equal prognostic value' analysis. It acknowledged that the assumption of equal prognostic value across the tests was not an unreasonable approach given the lack of evidence and the similarity in technical performance of the tests. The committee noted that, in the base case, the difference in total costs between the most expensive and least expensive test strategy was small. It also noted that, in the sensitivity analysis, although the difference in total costs between the different test strategies increased, it still remained relatively small and that the difference in total QALYs was also low. The committee concluded that, on balance, the cost effectiveness of the different tests and test strategies for EGFR‑TK mutation testing are likely to be similar.

The committee noted that, for some of the tests (high-resolution melt analysis, pyrosequencing combined with fragment length analysis and single-strand conformation polymorphism analysis), the 'equal prognostic value' analysis was the only economic modelling performed, and that test accuracy and clinical effectiveness were not assessed by the external assessment group because data were not available. The committee acknowledged that this represented a weaker evidence base than that for the therascreen EGFR PCR Kit, the cobas EGFR Mutation Test and Sanger sequencing tests. It noted further that 2 tests (therascreen EGFR Pyro Kit and next-generation sequencing) were not included in the assessment because no information on failure rates in clinical practice in the UK was available. The committee acknowledged that, although the cost effectiveness of the different tests and test strategies for EGFR‑TK mutation testing are likely to be similar, there is insufficient evidence to support this conclusion.

The committee noted that the analysis in NICE's technology appraisal guidance on gefitinib for the first-line treatment of locally advanced or metastatic non-small-cell lung cancer was primarily based on data from the IPASS trial, which used the therascreen EGFR PCR Kit to classify tumours of patients as EGFR‑TK mutation positive or negative. The committee acknowledged that the recommendation of gefitinib for the first-line treatment of locally advanced or metastatic NSCLC in patients whose tumours test positive for the EGFR‑TK mutation (NICE's technology appraisal guidance on gefitinib) implies that the therascreen EGFR PCR Kit is recommended and cost effective as part of the test-treat strategy. The committee concluded that, for the cobas EGFR Mutation Test and for Sanger sequencing-based methods, an equivalent evidence base exists, and therefore these tests and the therascreen EGFR PCR Kit can be considered clinically effective and cost effective for informing first-line treatment decisions in patients with previously untreated, locally advanced or metastatic NSCLC in the NHS. The committee further concluded that, for the non-Sanger sequencing-based tests (high-resolution melt analysis, pyrosequencing combined with fragment length analysis and single-strand conformation polymorphism analysis) and for tests not included in the external assessment group's assessment (the therascreen EGFR Pyro Kit and next-generation sequencing), the evidence was insufficient to allow any recommendations to be made on their use.