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

How outcomes were assessed

5.1 The assessment was performed by an External Assessment Group and consisted of a systematic review, a web-based survey and the development of a decision analytic model.

5.2 The systematic review was carried out to identify evidence on the technical performance and clinical effectiveness of the different options available to detect epidermal growth factor receptor tyrosine kinase (EGFR‑TK) mutations in previously untreated locally advanced or metastatic non-small-cell lung cancer (NSCLC), and so adults who may benefit from first-line treatment with EGFR‑TK inhibitors.

5.3 The web-based survey was conducted to gather data on the technical performance characteristics and costs of EGFR‑TK mutation tests in use in NHS laboratories.

5.4 A decision analytic model was developed to assess the cost effectiveness of different methods of EGFR‑TK mutation testing in helping to decide between treatment with standard chemotherapy and EGFR‑TK inhibitors for patients with locally advanced or metastatic NSCLC. Three different analytic approaches, described below, were used to calculate cost effectiveness, each involving different levels of evidence.

Technical performance

5.5 One study identified from the systematic review evaluated the technical performance of EGFR‑TK mutation tests. The study was conducted in the Department of Molecular Diagnostics at the Royal Marsden Hospital and the Institute of Cancer Research. The study reported data for 2 years of EGFR‑TK mutation testing from January 2009 to January 2011. During year 1 of the testing, the therascreen EGFR PCR Kit was used. During year 2, a combination of the therascreen EGFR PCR Kit, fragment analysis (for exon 19 deletions and exon 20 insertions) and Sanger sequencing (for the rarer exon 19 or exon 21 mutations) was used. A total of 121 patients were tested during year 1 and 755 during year 2. The mean turnaround time for the therascreen EGFR PCR test alone during year 1 was 4.9 business days (95% confidence interval [CI] 4.5 days to 5.5 days). However, the actual time from the test request to the result was 17.8 days (95% CI 16.4 days to 19.4 days). The total test failure rate for the first year of the study was 19% of all samples assessed, but this improved over time from a failure rate of 33% over the first 3 months to 13% during the last 3 months of year 1 testing. The total failure rate was lower in the second year of the study at only 5% of all samples assessed.

5.6 There were 24 UK laboratories participating in the 2012–2013 UK National External Quality Assessment Service (NEQAS) pilot scheme for EGFR‑TK mutation testing. Of these, 14 provided information to NICE during the scoping phase of the assessment and were invited to participate in the survey. Thirteen of the 14 laboratories completed the web-based survey.

5.7 The therascreen EGFR PCR Kit was the most commonly used EGFR‑TK mutation test, with 6 laboratories using it. A combination of fragment length analysis and pyrosequencing was used in 2 laboratories. Sanger sequencing was used in 2 laboratories. However, one of these laboratories also uses the cobas EGFR Mutation Test for verifying mutations or when the sample contains insufficient tumour cells for Sanger sequencing (less than 30%). The second of these laboratories also uses fragment length analysis and real-time PCR to follow up samples found to be negative with Sanger sequencing. Single-strand conformation analysis, high-resolution melt analysis and pyrosequencing were used in single laboratories. One laboratory also provided information on a next-generation sequencing method that is being developed and validated.

5.8 The survey results showed that there were no clear differences between tests. The number of samples screened for EGFR‑TK mutations in a typical week varied by laboratory from less than 5 (6 laboratories) to more than 20 (3 laboratories). The frequency at which the laboratories ran the tests ranged from daily to every other week. Batch sizes ranged from less than 3 samples to 10 samples but most laboratories stated that they would match demand rather than waiting for a minimum batch size.

5.9 Most laboratories had a turnaround time from receiving the sample to reporting the result to the clinician of 3–5 days or 6–7 days, with 1 laboratory reporting a turnaround of 24–28 hours (therascreen EGFR PCR Kit) and 1 laboratory reporting a turnaround of 8–10 days (therascreen EGFR PCR Kit). The estimated total number of failed samples ranged from 0% to 10%, with the number of failed samples because of insufficient tumour cells ranging from 0–5%. The most common reasons for failed tests were insufficient tumour cell count and poor-quality DNA or DNA degradation.

5.10 The cost of the EGFR‑TK mutation tests ranged from £110 to £190 and the price that the laboratories charged for the tests ranged from £120 to £200. When there was a difference between the test cost and the price charged, this ranged from £10 to £37.50 per test. No single test appeared to be more or less expensive than any of the other tests.

5.11 It was noted by UK NEQAS that error rates seen in the quality assurance scheme for EGFR‑TK mutation testing are not always method related, and may be because of processing and reporting problems. In addition, UK NEQAS noted that there had been no correlation between any method used for EGFR‑TK mutation testing and errors since the scheme was started in 2010.

Accuracy

5.12 Two randomised controlled trials and 4 cohort studies provided data on the accuracy of EGFR‑TK mutation testing for predicting the response to treatment with EGFR‑TK inhibitors in patients with advanced or metastatic NSCLC. Three studies included patients treated with gefitinib and 3 included patients treated with erlotinib.

5.13 Patient characteristics varied across studies. One study included mainly white patients and 1 study included mainly East Asian patients (4 studies did not report the ethnicity of patients). All studies reported that a high proportion of patients had metastatic disease. Most patients had a histological diagnosis of adenocarcinoma (45–100%), but 2 studies included some patients with squamous cell carcinoma (9–15%). Four studies mainly, or only, included patients who had never smoked, whereas 2 studies mainly included patients who were current or former smokers.

5.14 Five studies evaluated Sanger sequencing methods for identifying any EGFR‑TK mutation; 3 assessed exons 18 to 21, 1 assessed exons 19 to 21, and 1 assessed exons 18 to 24 (Sanger sequencing or WAVE-HS for inadequate samples [less than 50% tumour cells]). One study assessed the therascreen EGFR PCR Kit (the version designed to detect 29 mutations, including T790M).

5.15 The therascreen EGFR PCR Kit appears to have the best overall performance for discriminating between patients who are likely to benefit from EGFR‑TK inhibitor treatment and patients who are not. The sensitivity and specificity estimates using objective response as the reference standard were 99% (95% CI 94% to 100%) and 69% (95% CI 60% to 77%) respectively.

5.16 Of the 5 studies that used Sanger sequencing methods to identify EGFR‑TK mutations, 4 reported high estimates of specificity (more than 80%) and sensitivities ranged from 60% to 80% when objective response was used as the reference standard. The remaining Sanger sequencing study reported low specificity (61%) with high sensitivity (84%) for objective response as the reference standard.

Clinical effectiveness

5.17 Five randomised controlled trials provided data on the clinical effectiveness of EGFR‑TK inhibitors compared with standard chemotherapy in patients with advanced or metastatic NSCLC whose tumours tested positive for EGFR‑TK mutations. One additional study reported data for a subgroup of patients from the EURTAC trial whose samples had been re-analysed using a different EGFR‑TK mutation testing method. Three of the trials included only patients with EGFR‑TK mutation-positive tumours, and the remaining 2 trials (IPASS and First-SIGNAL) included all patients regardless of EGFR‑TK mutation status, but also reported a subgroup analysis for patients whose tumours tested positive for EGFR‑TK mutations. The trials compared the EGFR‑TK inhibitors gefitinib or erlotinib with various single-agent or combination standard chemotherapy regimens.

5.18 Patient characteristics varied across studies. Four studies were conducted in East Asia and 1 was conducted in Western Europe. One study included patients who had never smoked, 1 study included mainly patients who had never smoked (94%) and the rest included between 62% and 71% of patients who had never smoked. One study included only patients with a diagnosis of adenocarcinoma, whereas in the remaining studies approximately 90% had a diagnosis of adenocarcinoma. Most patients (more than 75%) in all studies had metastatic disease.

5.19 Two studies used Sanger sequencing methods to assess EGFR‑TK mutation status, but both limited the definition of positive EGFR‑TK mutation status to the presence of an 'activating mutation' (exon 19 deletions or exon 21 mutation L858R). The remaining studies used EGFR‑TK mutation tests that targeted a wider range of mutations. One study reported the results of a re-analysis of samples from the EURTAC trial using the cobas EGFR Mutation Test. The other study (IPASS) used the therascreen EGFR PCR Kit (the version designed to detect 29 mutations, including T790M). The North East Japan Study Group (NEJSG) trial used fragment length analysis, targeting exon 19 deletions, exon 21 point mutations (L858R, L861Q), exon 18 point mutations (G719A, G719C, G719S), and exon 20 point mutation (T790M). The First-SIGNAL trial used Sanger sequencing of exons 19 to 21.

5.20 All studies reported improvements in objective response, measured as relative risk. Objective response ranged from a relative risk of 1.51 (95% CI 1.23 to 1.88) to 3.89 (95% CI 2.34 to 6.68) for patients with EGFR‑TK mutation-positive tumours who were given EGFR‑TK inhibitors compared with patients given standard chemotherapy. All studies also reported statistically significant improvements or trends towards improvement in progression-free survival, with hazard ratios ranging from 0.16 (95% CI 0.10 to 0.26) to 0.54 (95% CI 0.27 to 1.10) for patients with EGFR‑TK mutation-positive tumours who were given EGFR‑TK inhibitors compared with patients given standard chemotherapy. Four studies reported overall survival but none found a statistically significant difference between patients given EGFR‑TK inhibitors and patients given standard chemotherapy, with hazard ratios ranging from 0.89 (95% CI 0.63 to 1.24) to 1.04 (95% CI 0.65 to 1.68).

5.21 The results from the IPASS trial showed that progression-free survival in patients with EGFR‑TK mutation-negative tumours was statistically significantly shorter when patients were treated with EGFR‑TK inhibitors than with standard chemotherapy (hazard ratio [HR] 2.85, 95% CI 2.05 to 3.98). A similar trend for patients with EGFR‑TK mutation-negative tumours, although not statistically significant, was observed in the First-SIGNAL trial (HR 1.42, 95% CI 0.82 to 2.47).

Cost effectiveness

5.22 The External Assessment Group received the health economic model submitted by AstraZeneca for NICE technology appraisal guidance 192 (Gefitinib for the first-line treatment of locally advanced or metastatic non-small-cell lung cancer). The External Assessment Group also took into account amendments and corrections to the model that were accepted by the appraisal committee for NICE technology appraisal guidance 192. This model calculates the expected cost effectiveness of gefitinib compared with standard chemotherapy for the first-line treatment of locally advanced or metastatic NSCLC in patients with a positive EGFR‑TK mutation test based on the therascreen EGFR PCR Kit. The External Assessment Group used the AstraZeneca model to develop a de novo model that included patients with a positive, negative or unknown EGFR‑TK mutation test result.

5.23 The External Assessment Group developed a decision tree and a Markov model to analyse the long-term consequences of technical performance and accuracy of the different EGFR‑TK mutation tests and test combinations followed by treatment with either standard chemotherapy or an EGFR‑TK inhibitor in patients with NSCLC. The decision tree was used to model the test result (positive, unknown or negative) and the treatment decision. Patients with a positive test result receive an EGFR‑TK inhibitor. Patients with a negative test result or an unknown EGFR‑TK mutation status receive standard chemotherapy (pemetrexed and cisplatin). The Markov model was used to estimate the long-term consequences in terms of costs and quality-adjusted life years (QALYs). The model has a cycle time of 21 days (resembling the duration of 1 cycle of chemotherapy), and a time horizon of 6 years. In the model, after a treatment decision is made, patients can either have progression-free disease (subdivided into 'response' and 'stable disease' based on objective response rate), experience disease progression or die.

5.24 The proportions of positive and negative EGFR‑TK mutation test results were based on: the estimated proportions of patients with NSCLC and EGFR‑TK mutation-positive tumours in England and Wales (16.6%, standard error 0.8%); the test accuracy (sensitivity and specificity with objective response to EGFR‑TK inhibitor as reference standard); and the proportion of patients with an unknown test result, based on data from published studies (IPASS and Jackman et al. 2007). The proportions of positive, negative and unknown EGFR‑TK mutation test results for the therascreen EGFR PCR Kit were 32.8%, 44.6% and 22.7% respectively. The proportions of positive, negative and unknown EGFR‑TK mutation test results for Sanger sequencing of exons 18 to 21 were 29.0%, 33.4% and 37.7% respectively. In the 'assumption of equal prognostic value' analysis, the proportions of positive, negative and unknown EGFR‑TK mutation test results were assumed equal to the therascreen EGFR PCR Kit for all tests and test strategies.

5.25 The objective response rates were based on data from published studies (IPASS, First-SIGNAL and Yang et al. 2008). For EGFR‑TK mutation-negative or -unknown tumours (treated with standard chemotherapy), objective response rates were adjusted to correspond to treatment with pemetrexed and cisplatin. Objective response rates for EGFR‑TK mutation-positive, -negative and ‑unknown tumours identified using the therascreen EGFR PCR Kit were 0.712, 0.335 and 0.403 respectively. Objective response rates for EGFR‑TK mutation-positive, -negative and -unknown tumours identified using Sanger sequencing of exons 18 to 21 were 0.731, 0.604 and 0.403 respectively. In the 'assumption of equal prognostic value' analysis, the objective response rates for EGFR‑TK mutation-positive, -negative and -unknown tumours were assumed equal to the therascreen EGFR PCR Kit for all tests and test strategies.

5.26 Progression-free survival and overall survival after testing with the therascreen EGFR PCR Kit were modelled using Weibull regression models based on the IPASS trial and a hazard ratio favouring treatment with an EGFR‑TK inhibitor (HR 0.43, 95% CI 0.34 to 0.53). For testing using Sanger sequencing of exons 19 to 21, progression-free survival and overall survival for patients with EGFR‑TK mutation-positive or -negative tumours were modelled using Kaplan-Meier curves extracted from the First-SIGNAL trial. Progression-free survival and overall survival for patients with tumours of unknown EGFR‑TK mutation status were based on the IPASS Weibull model for unknown mutations. For testing using Sanger sequencing of exons 18 to 21, progression-free survival and overall survival were assumed equal to testing using Sanger sequencing of exons 19 to 21.

5.27 The test costs were based on the prices charged by the NHS laboratories in England and Wales involved in the web-based survey (see table 1). In the case of an unknown EGFR‑TK mutation status, no test costs were taken into account if there was a pre-laboratory clinical failure, but full test costs were taken into account if there was a technical failure in the laboratory.

Table 1 EGFR‑TK mutation test costs

5.28 Results from the 'comparative effectiveness' analysis showed the therascreen EGFR PCR Kit to be both less effective and less costly compared with Sanger sequencing of exons 19 to 21, with an incremental cost-effectiveness ratio (ICER) of £32,167 saved per QALY lost. Adjustments to costs and the proportions of patients with unknown mutation status in sensitivity analyses had little effect on the results. When treatment costs and adverse event costs were updated to 2012 costs, the ICER was £32,196 saved per QALY lost for the therascreen EGFR PCR Kit compared with Sanger sequencing. When the proportions of patients with unknown mutation status were based on the results from the web-based survey rather than information from published trials, the ICER was £34,555 saved per QALY lost for the therascreen EGFR PCR Kit compared with Sanger sequencing.

5.29 The External Assessment Group explained that the lower costs and QALYs for the therascreen EGFR PCR Kit were because patients with EGFR‑TK mutation-negative tumours had shorter overall survival in the IPASS trial (therascreen EGFR PCR Kit) than in the First-SIGNAL trial (Sanger sequencing of exons 19 to 21), whereas the outcome was comparable for patients whose tumours were EGFR‑TK mutation positive. For patients whose tumours were EGFR‑TK mutation unknown, overall survival was the same by assumption. Therefore, on average, with the therascreen EGFR PCR Kit patients had shorter overall survival, resulting in fewer QALYs and reduced costs compared with Sanger sequencing of exons 19 to 21.

5.30 The External Assessment Group noted that this analysis is particularly problematic because of the assumption that the differences in relative treatment response, progression-free survival and overall survival between the results of the First-SIGNAL trial (Sanger sequencing of exons 19 to 21) and the results of the IPASS trial (therascreen EGFR PCR Kit) were solely because of the different EGFR‑TK mutation tests used to distinguish between patients whose tumours were EGFR‑TK mutation positive (and receive EGFR‑TK inhibitor treatment) and patients whose tumours were EGFR‑TK mutation negative (and receive standard chemotherapy).

5.31 Results from the 'linked evidence' analysis also showed the therascreen EGFR PCR Kit to be both less effective and less costly than Sanger sequencing of exons 18 to 21 at an ICER of £31,849 saved per QALY lost. Sensitivity analyses had little effect on the results. When the treatment costs and adverse event costs were updated to 2012 costs, the ICER was £34,169 saved per QALY lost for the therascreen EGFR PCR Kit compared with Sanger sequencing of exons 18 to 21. When the proportions of patients with unknown mutation status were based on the results from the web-based survey rather than information from published trials, the ICER was £31,880 saved per QALY lost for the therascreen EGFR PCR Kit compared with Sanger sequencing of exons 18 to 21.

5.32 The reason for the lower costs and QALYs for the therascreen EGFR PCR Kit were the same as for the 'comparative effectiveness' analysis, as described in section 5.28.

5.33 In addition to the assumption described in section 5.30, the 'linked evidence' analysis also assumed that the relative progression-free survival and overall survival for Sanger sequencing of exons 18 to 21 correlated perfectly with the relative progression-free survival and overall survival for Sanger sequencing of exons 19 to 21.

5.34 In the 'assumption of equal prognostic value' analysis, the comparative effectiveness, test accuracy and proportion of patients with unknown mutation status for each test strategy were assumed equal to those of the therascreen EGFR PCR Kit. Therefore, the test strategies only differed with respect to costs. Results showed that the test strategy of Sanger sequencing or the cobas EGFR Mutation Test for samples with insufficient tumour cells was the least expensive (£15 [0.06%] cheaper than Sanger sequencing of exons 18 to 21 alone), and fragment length analysis combined with pyrosequencing was the most expensive strategy (£33 [0.13%] more expensive than Sanger sequencing of exons 18 to 21 alone).

5.35 The External Assessment Group did not include next-generation sequencing and the therascreen EGFR Pyro Kit in any of the cost-effectiveness analyses because of a lack of data. No published studies were identified for either of these methods and neither method is currently in routine clinical use in any NHS laboratories in England and Wales.