Evidence reviews for genetic testing for melanoma

1.1.4.1. Included studies

A systematic literature search was conducted for this review on genetic testing for people with melanoma. This returned 5,231 references (see appendix B for the literature search strategy). Based on title and abstract screening against the review protocol, 5,150 references were excluded, and 81 references were ordered for screening based on their full texts.

Of the 81 references screened as full texts, 3 references met the inclusion criteria specified in the review protocol for this question (appendix A) and were specific to people with stage 2C-3 melanoma. Following discussion with the committee it was agreed that the inclusion criteria should be expanded to all people with melanoma (regardless of stage), increasing the final number of references to 13. The clinical evidence study selection is presented as a diagram in appendix C.

Re-run searches identified an additional 1 reference for inclusion.

1.1.4.2. Excluded studies

See Appendix I for a list of references for excluded studies, with reasons for exclusion.

1.1.7.1. Included studies

A single search was performed to identify published economic evaluations of relevance to any of the questions in this guideline update (see appendix B). This search retrieved 7,545 studies. Based on title and abstract screening, 7,538 of the studies could confidently be excluded for this question. 7 studies were excluded following the full-text review. Thus, the review for this question does not include any study from the existing literature.

1.1.7.2. Excluded studies

See Appendix I for excluded studies and reasons for exclusion.

1.1.10.1. The outcomes that matter most

The committee agreed that that both sensitivity/specificity and likelihood ratios were suitable methods of visualising the diagnostic accuracy data and that the quality assessment should be done on the likelihood ratios due to the existence of an established interpretation of these values, making it easier to assess imprecision.

IHC has the benefit of being conducted very quickly compared to standard tests for BRAF mutations, such as COBAS, and there is the potential for IHC to be used to detect BRAF mutations when immediate treatment is required and an urgent test is required, such as for people with fast progressing disease. IHC is deemed to produce very few false positive results and therefore a positive test could be used to diagnose BRAF mutation without further testing. As IHC only detects V600e mutations, a negative test would always require further testing to determine BRAF mutation status. Based on this, the committee agreed that specificity and positive likelihood ratios were the most important outcomes when assessing immunohistochemistry.

The committee agreed that it was appropriate to assess the accuracy of IHC for detecting all BRAF mutations and for detecting specifically v600e mutations. The former approach would reflect the accuracy of IHC when used in practice, and the latter would reflect the accuracy of IHC for what it was designed to do as IHC only aims to detect v600e mutations.

A false positive would result in a person being incorrectly staged and being classified as BRAF mutant. This may lead to them receiving targeted treatment instead of a more suitable therapy, such as adjuvant pembrolizumab or other immunotherapies. False positive patients will not respond to targeted treatment in the expected manner as they do not possess the BRAF mutation, and will eventually need to switch to another treatment, potentially after experiencing disease progression.

A false negative would not have significant downstream consequences. The person would go on to receive the previous standard of care – BRAF analysis with a COBAS test – to confirm or exclude BRAF mutation.

A true positive result would result in the person being correctly upstaged, classified as BRAF mutant and becoming eligible for additional treatment options.

A true negative result would result in the person being correctly classified as BRAF wild-type and their staging would be unaffected.

1.1.10.2. The quality of the evidence

All evidence came from retrospective cohort studies that were directly applicable to the review question. Studies were typically of low risk of bias. Areas in which there was a risk of bias stemmed from a lack of blinding or the use of composite reference standards, in which a person’s true BRAF status was determined by one of numerous possible tests, allowing different samples to undergo different reference standards.

The committee advised that in clinical practice IHC would be used to rule-in people with a BRAF mutation, with a positive result classifying someone as BRAF-positive and a negative result requiring that the person undergo subsequent testing with COBAS or an alternative genomic BRAF test.

Studies reported a variety of different reference standards. The committee advised that it was important to look at studies using COBAS as the reference standard (either COBAS alone, or COBAS with subsequent testing for discrepant cases between COBAS and IHC) as in practice, decisions about subsequent treatment are based on the results of the COBAS test alone.

However, the committee agreed that as COBAS also has the potential for false negative and false positive results, there is a risk that this would lead to an inaccurate measure of the diagnostic accuracy of IHC and that the reference standard of NGS is preferable as this is a true gold standard test that would allow a comparison between IHC and the person’s actual mutation status. It was agreed that the economic model would primarily use data where diagnostic accuracy was assessed using NGS as a reference standard. They agreed that studies using COBAS as a reference standard was still useful as it would give an indication of what would happen when the tests are used sequentially.

When using COBAS alone as a reference standard, there was a high degree of heterogeneity between studies in their reported positive likelihood ratios. Heterogeneity was still present when using NGS as the reference standard although it was less pronounced.

1.1.10.3. Benefits and harms

The committee advised that evidence suggests that people with clinical stages IIA-C have similar 5- and 10-year mortality rates, comparable to those with stage IIIA-B melanoma. People with stage IIC are at a particularly high risk of mortality, with evidence suggesting 5-and 10- year mortality rates slightly higher than those with stage IIIB melanoma. The committee highlighted challenges with retrieval of genetic samples from storage and that it was more practical to test for BRAF status at the point of diagnosis rather than when they would become eligible for targeted therapy, e.g. upon progression, as delays to treatment due to sample retrieval and testing time can cause harms to patients who may be at risk of further deterioration. The committee also wished to recommend that BRAF analysis of melanoma tissue samples be arranged by the local MDT in order to provide a more coordinated process. The pathology report on the primary lesion should also include the relevant tissue block suitable for molecular genetic testing, as determined by the dermatopathologist within the MDT.

The committee agreed that BRAF testing is essential to identify whether people are eligible for targeted therapies and, although recommendations made in evidence review F deprioritise use of these treatments in people with unresectable stage III or IV, there is still a significant portion of people who would receive targeted therapy as first line treatment if they were at risk of rapid progression, preferred to use targeted therapy after consideration of the safety profile compared with immunotherapy agents, or who would switch to targeted therapies after immunotherapy. Additionally, targeted therapies are used in adjuvant settings for people with resectable stage III melanoma.

The committee also agreed that although people with stage IIA-C disease would not immediately benefit from having their BRAF status known, testing these people at the point of diagnosis has practical utility. A significant portion of people with stage IIA-C disease will relapse and having their BRAF status already known will speed up decisions surrounding which treatment to give.

Based on this evidence the committee agreed to recommend that BRAF testing be offered to people with clinical stage IIC-IV melanoma and be considered for people with clinical stage IIA or IIB melanoma.

The committee agreed to keep recommendations that BRAF analysis not be used in people with stage IA-IB melanoma due to the low risk of BRAF mutation and better prognosis in these groups of people.

The committee agreed that false positive results with IHC are very rare and that the diagnostic accuracy evidence confirms this, due to the high specificity. As IHC only detected the V600e BRAF mutation, many people with a (non-V600e) BRAF mutation will be missed if they were to be tested by IHC alone. This is reflected in the evidence for the sensitivity of IHC, which is smaller in comparison to its specificity. Although V600e is the most common form of BRAF mutation in people with melanoma, other variants (particularly V600k) are also common. Economic modelling estimated that IHC followed by PCR Cobas is more expensive than PCR Cobas alone, but results in a greater number of people appropriately receiving targeted therapy.

The committee agreed that the faster speed in which IHC can be processed (in hours instead of days/ weeks) is a clear clinical benefit to IHC. This is particularly important in people with poorer prognosis (such as people with metastatic cancer) who require rapid treatment. They also agreed that after a sample is taken, there are difficulties establishing the stage of disease.

As such, the committee agreed that IHC be considered as the first test for samples undergoing BRAF analysis but that negative tests should go on to receive confirmatory testing using an alternative BRAF genomic test.

The committee were aware that availability of the necessary equipment and technical expertise to analyse and interpret IHC assays will vary between centres and that particularly for smaller centres who only administer a small number of BRAF tests, it may not be cost effective or feasible for them to purchase the necessary equipment. They accounted for this possibility in the recommendations. Additionally, the committee agreed that decisions to test for BRAF mutations should take into account suitability for targeted or systemic therapy if they were to test positive.

1.1.10.4. Cost effectiveness and resource use

No published economic evidence was identified from the systematic review. However, the committee was presented with economic evidence from a de novo cost consequence analysis developed for the guideline. The model assessed the costs and effectives of different approaches to genetic testing at diagnosis for identifying BRAF mutations in patients with stage IIC and III melanoma. The testing approaches compared were PCR Cobas alone versus using upfront immunohistochemistry (IHC) with PCR Cobas reserved for only those patients who test negative with IHC.

In advising on an appropriate structure for the model, the committee noted that the greatest benefit of genetic testing with IHC is the reduced test turnaround time compared to PCR Cobas (e.g., same day result vs a waiting period of 14 days), which avoids delays in patients receiving adjuvant targeted therapies or systemic targeted therapies on recurrence. However, IHC is limited in its ability to detect all BRAF mutations and can only identify patients with BRAF V600E mutations. This means that if one were to test with only IHC, a number of patients with other actionable BRAF mutations (e.g., BRAF V600K, V600R, V600D, V600M) would be incorrectly identified as BRAF wildtype. Thus, anyone who tests negative with IHC should then be tested using a secondary genetic test, such as PCR Cobas, which can identify all relevant and actionable BRAF mutations. To capture the negative consequences of a longer test turnaround time, the model was structured so that anyone who receives a PCR Cobas test can either get a test result or die before receiving a test result. The probability of death during the longer test turnaround time although small, potentially has an impact on the number of patients that can ultimately go on to receive targeted therapy. Additionally, the committee was interested in the costs of each testing approach as well as the outcome of the number of patients who go on to appropriately receive targeted therapy as a result of being correctly identified as having a BRAF mutation when using each testing approach.

The committee was presented with the base case model results for two distinct populations, patients with stage IIC melanoma and patients with stage III melanoma. Two different models were built for these two distinct populations as the current treatment pathway for stage IIC is different than stage III melanoma. Currently, those with stage IIC melanoma are only eligible for adjuvant therapy on recurrence, however many of those with resectable stage III melanoma are immediately eligible for adjuvant targeted therapy at diagnosis. The committee also noted that clinical trials are currently ongoing in which could change the pathway of care such that those with stage IIC melanoma would eventually also become eligible for adjuvant therapy immediately at diagnosis rather than only on recurrence. In both patient populations, IHC followed by PCR Cobas is more expensive than PCR Cobas alone, but results in a greater number of people appropriately receiving targeted therapy.

The committee was also presented the results of several deterministic sensitivity analyses, in which the results of the analysis remained largely robust to a range of scenarios when varying any of the model’s input parameters within the range of their uncertainty. One parameter that had the greatest effect on the results was the cost of IHC. Probabilistic sensitivity analysis provided congruent results to the base case analysis. The outputs of the probabilistic analysis also provided further support of the model results as there were no iterations for either of the patient population in which the testing approach using IHC with PCR Cobas was associated with worse outcomes compared to PCR Cobas alone.

The committee therefore discussed the appropriateness of the costings used in the model, which relied on data from two micro-costing studies, and agreed that the base case cost of IHC used in the model was likely too low. The micro-costing for IHC in the model was in part based on an IHC micro-costing paper for detection of another mutation. Although the process for IHC would be the same, and therefore our estimates of staff time would be comparable, different antibodies would be needed for BRAF testing. The committee noted the antibody needed for BRAF testing is likely to cost thousands of pounds, which was underestimated in the model at only a few hundred pounds. Individual committee members with knowledge of IHC labs indicated that the cost per IHC test to detect a BRAF V600E mutation may range from £40-£200. However, the higher estimate of £200 was estimated based on validation costs that would be required to set up the IHC platform and would only be incurred within the first year of implementing the test. Therefore, the committee felt that the actual ongoing cost per IHC test would likely to be less than £200. Considering the results of the threshold analyses, the committee felt confident that a testing approach using IHC with PCR Cobas compared to PCR Cobas alone would be highly likely to be cost-effective in patients with stage III melanoma. For patients with stage IIC melanoma, the committee acknowledged that if the actual cost per test of IHC was £200, which was very close to the upper value of £204 identified in the threshold analysis that a testing approach using IHC with PCR Cobas might not be cost-effective. However, as previously noted, the £200 figure for an IHC test included costs associated with validation, that if they were to be spread over the lifetime of the testing equipment (rather than solely allocated within the first year of testing) the cost per IHC test would be smaller and likely be in the range for the testing approach using IHC with PCR Cobas to be considered cost-effective. Additionally, the committee noted that several labs already have IHC testing capacity and therefore validation of the IHC test for the BRAF V600E mutation might not be as costly and the cost per IHC testing more likely to be in the range to be considered cost-effective (i.e., less than £204). Finally, the committee made note of the fact that one of the primary reasons for the high cost of BRAF V600E mutation testing with IHC is due to one of the antibodies required for the test being on patent. The committee noted that this antibody is due to come off patent in the next few years, which will likely further reduce the cost of testing with IHC increasing the cost-effectiveness of the testing approach.

In view of these considerations, the committee made a recommendation to consider IHC to be used as an initial screening test to identify BRAF V600E mutations alongside a secondary genetic test in those testing negative in patients with either stage IIC or III melanoma.

Although the scope of this review question was limited to those with stage IIC and III melanoma, the committee also made consider recommendations using the same testing approach in patients with stage IIA and IIB melanoma. This was justified based on survival data from the updated AJCC 2018 staging criteria (Gershenwald et al. 2017), where these two populations had similar survival to that of the stage IIC (and IIIA) patients and therefore likely to have similar rates of recurrence. The committee believed that testing may have value in these patients as it would allow them rapid access to therapy upon progression, and that a delay in treatment at this point would be associated with harm if they were rapidly progressing. However, given the uncertainty in making this extrapolation the committee felt it would be better for the recommendations to only consider an IHC with PCR Cobas testing approach, thereby giving the clinician the option to pursue such testing if they thought it would be of use to the patient.

The committee also felt it was important not to mandate the use of IHC with PCR Cobas, as they worried this would result in labs without this testing capacity sending IHC tests elsewhere, which would both increase the costs of this testing approach and the test turnaround time, thereby negating one of the greatest benefits of this testing approach. Therefore, the recommendation indicated that IHC could be used when available, but if it was not available, another genetic test could be used alone.

The committee also considered the potential resource impact of these recommendations. For centres that already have IHC equipment available and already using it as a testing method, the committee noted the recommendations would have only have a small impact on resource use as such centres would only need to purchase the appropriate antibodies for the IHC BRAF test. For centres that do not have IHC equipment, the committee noted that this would increase resource use, both in the form of the upfront costs required to set up IHC testing, validation of the testing approach in the first year, and ensuring staff are trained such that they are skilled enough to appropriately interpret the test. However, on balance, the committee felt that this increase in resource use was not likely to be prohibitively expensive, would be limited to the first year of implementation, and would likely decrease in the future as the required IHC antibodies come off patent. While the committee expressed some uncertainty over the cost, in light of the results of the economic model and their clinical knowledge, the committee agreed these costs were likely to remain small and come with a number of beneficial outcomes including a quicker time to result for those testing positive with IHC and a greater number of people appropriately receiving targeted therapy.