Testing strategies for Lynch syndrome in people with endometrial cancer

4.1 A patient expert highlighted that identifying Lynch syndrome after a diagnosis of endometrial cancer means interventions and surveillance can be adopted to reduce the risk of other Lynch syndrome-associated cancers or detect them earlier. Another benefit is that it allows testing for the condition to be offered to relatives, who can be identified as having Lynch syndrome before they have cancer. The patient expert commented that if a person knows they have Lynch syndrome they can make lifestyle changes to help reduce their cancer risk. As well as offering clinical surveillance to people with Lynch syndrome, such as colonoscopies, the symptoms of Lynch syndrome-associated cancers can be highlighted to make sure people seek medical advice if they have symptoms. Knowing that they are at higher risk of gynaecological cancer may also help people make decisions about family planning. The committee noted that endometrial cancer is often the first Lynch syndrome-associated cancer that people are diagnosed with. So at this point there is an opportunity to identify Lynch syndrome before other associated cancers, such as colorectal cancer, develop. The committee concluded that people and their families would likely benefit substantially if Lynch syndrome was identified after endometrial cancer is diagnosed.

4.2 There can be considerable anxiety and uncertainty associated with genetic testing for hereditary cancer syndromes such as Lynch syndrome. A patient expert explained that waiting for test results can be a very anxious time. Test results can have a substantial effect on a person, so it is very important that people understand the full implications of a diagnosis of Lynch syndrome, for themselves and their families. This is especially important for people with a learning disability, who may need support from a carer to fully engage in discussions about testing and to give informed consent. For people who have not had children yet, or who want more children, there may be anxiety and uncertainty about risk-reducing surgery, because after this it is not possible to give birth. Making this decision needs a good understanding of their risk of cancer (for example, based on their specific pathogenic variant of a mismatch repair [MMR] gene). People may also have concerns about the invasive nature of surveillance for cancer. The committee concluded that genetic counselling is very important for people with Lynch syndrome, or who are at risk of having Lynch syndrome, because it can help people understand if genetic testing is appropriate or not. It can also help people understand the importance of telling extended family about their risk of having Lynch syndrome and the benefits of being tested. The committee also noted the importance of developing information resources to help people decide about testing, and the need for these to be available at the appropriate time.

4.3 A patient expert explained that waiting for test results can be a very anxious time. The strategies assessed included immunohistochemistry (IHC) and microsatellite instability (MSI) as the first tests used to identify tumours with MMR deficiency. The external assessment group (EAG) commented that it did not find data on the time taken to get test results when using IHC or MSI. However, clinical experts commented that based on their experience in the NHS the time taken to get IHC results (about 1 day) is much shorter than for MSI results (about 6 weeks). They highlighted that when MSI testing is used, information may not be available for the first multidisciplinary team discussions on a person's cancer where decisions about treatment and further testing are made. They also noted that there is a risk that results arriving weeks later, separate to the histopathology report, may be missed and not acted on. Also, if a person's tumour is quickly identified as having MMR deficiency, they are able to be considered for clinical trials for new treatments such as immunotherapies. Clinical experts highlighted that it is important for everyone with cancer to have the opportunity to enter relevant clinical trials, and that this is a way to access new treatments which could benefit them. The committee noted a consultation comment that increased genomic analysis of tumours and streamlining of the testing pipeline could mean that MSI test results based on next generation sequencing technology will be available in a shorter time. However, the committee noted that this is a future development and still may not mean that MSI test results are available as quickly as IHC test results. Clinical experts also commented that an advantage of IHC testing is that it shows which MMR gene is likely to contain a pathogenic mutation. The committee concluded that there may be benefits for people and their clinicians from using IHC compared with MSI as a first test for potential Lynch syndrome.

4.4 The EAG commented that in studies assessing both MSI and IHC testing on the same sample of people (4 studies, see section 3.12) there was no statistically significant difference between the accuracy of the tests. But the committee noted that the studies were unlikely to be powered to show any difference. The committee preferred to use comparative estimates of sensitivity and specificity for IHC and MSI from the individual studies, rather than ranges of midpoint estimates from all 4 studies. It noted that sensitivity estimates were generally higher for IHC than for MSI (that is, using IHC as the first test may detect more people with Lynch syndrome). Clinical experts commented that a concern about MSI testing is that it may miss tumours with mutations in MSH6 and they highlighted the relatively high prevalence of mutations in this gene in Lynch syndrome-associated endometrial cancer (see section 3.6). The committee further noted that the level of concordance between IHC and MSI studies was in a range of 68% to 100% (see section 3.13). The committee concluded that there was uncertainty about the relative accuracy of IHC and MSI testing for potential Lynch syndrome done on endometrial tumour samples, but there was some evidence that IHC may detect more people with Lynch syndrome.

4.5 An unpublished manuscript of the PETALS study was given to the committee as academic in confidence. This study has now published as Ryan et al. (2020). The study authors commented that their results showed that IHC was more sensitive than MSI. The sensitivity of MSI testing was 56.3% compared with 100% for IHC testing. The MSI test used in PETALS was the Promega MSI analysis system v1.2, which assesses 5 mononucleotide repeat markers. The committee noted that this was in general agreement with the results of the 4 studies directly comparing IHC and MSI in the EAG's systematic review (see section 4.4). The study authors also commented that people with both MSI and IHC test results that did not indicate potential Lynch syndrome did not have the reference standard (germline testing) because of the cost. The committee noted that this may be a limitation of the study. It concluded that PETALS was likely to be highly relevant for this assessment. This was because it was a recent UK study that assessed IHC, MSI and MLH1 promoter hypermethylation testing in a population in the NHS who were newly diagnosed with endometrial cancer.

4.6 For the effect of colonoscopic surveillance on colorectal cancer incidence in the model, the EAG used a hazard ratio of 0.387 from Järvinen et al. (2000). The committee noted that this was an observational study and the true effect size was uncertain. The EAG did a scenario analysis in which no benefit of colonoscopy on colorectal cancer incidence was assumed. This was the only scenario in which the incremental cost-effectiveness ratios (ICERs) for all of the testing strategies compared with no testing were above £20,000 per quality-adjusted life year (QALY) gained, although the ICER for strategy 4 (IHC then MLH1 promoter methylation testing) was only just higher. Clinical experts commented that it would not be ethical to do a trial in which people with Lynch syndrome were randomised to have colonoscopic surveillance or not. They highlighted that the recent Guidelines for the management of hereditary colorectal cancer from the British Society of Gastroenterology (BSG)/Association of Coloproctology of Great Britain and Ireland (ACPGBI)/United Kingdom Cancer Genetics Group (UKCGG) recommend colonoscopic surveillance for people with Lynch syndrome. The committee concluded that although there is uncertainty about the effect of colonoscopic surveillance on colorectal cancer incidence, the EAG's scenario assuming no benefit was extreme and unlikely to be realistic.

4.7 Clinical experts commented that the extent of surveillance for gynaecological cancers offered to people with Lynch syndrome varies across the NHS. There was also considerable uncertainty about how effective this surveillance is in reducing the occurrence or severity of gynaecological cancer. The EAG included the cost and impact of gynaecological surveillance in its base-case model but highlighted that removing it in a scenario analysis did not have a large effect on results. Clinical and patient experts (see section 4.1) commented that, even if gynaecological surveillance is not done, raising awareness of the early symptoms of gynaecological cancers for people with Lynch syndrome is likely to improve early diagnosis.

4.8 Strategy 11 assessed the costs and benefits of going straight to germline testing for people diagnosed with endometrial cancer. That is, no first testing of tumour samples to identify people likely to have Lynch syndrome. Clinical experts commented that this would result in more variants of uncertain significance being detected (that is, mutations in MMR genes which may or may not cause Lynch syndrome). This is because initial tumour tests rule out non-pathogenic mutations (with no MMR deficiency in the tumour). This would mean staff time would be needed to analyse and assess if the person should be considered as having Lynch syndrome. Tests on endometrial tumour tissue may be needed to determine if the tumour was MMR-deficient. Costs related to this were not included in the economic model. In addition, clinical experts commented that if everyone with endometrial cancer had germline testing this would drastically increase the workload of genetic services, who would need to offer genetic counselling to ensure informed consent for the tests. Clinical experts also highlighted that there have been fewer studies done on Lynch syndrome-causing mutations for some ethnic groups, so there were likely to be more variants of uncertain significance identified by a straight to germline testing strategy (without the benefit of information from the tumour tests). The committee concluded that the costs of strategy 11 were likely to have been underestimated in the model.

4.9 In the economic model, a reduction in health-related quality of life for people with colorectal cancer was only assumed to happen at stage 4 of the disease. A relatively small reduction in health-related quality of life for people with endometrial cancer was also assumed, which lasted for only 1 year in the model. A scenario analysis in which people with stage 3 colorectal cancer also had reduced health-related quality of life increased the incremental QALY gain for all strategies compared with no testing (and cost effectiveness compared with no testing). The committee concluded that the effect of cancer on health-related quality of life was likely to have been underestimated in the model, so the cost effectiveness of testing for Lynch syndrome may also have been underestimated.

4.10 The EAG explained that the post-test clinic-related costs and follow-up costs it had used in its model (about £141) were for the time taken to give people their test results. Clinical experts commented that this cost was likely to be an underestimate. The committee also discussed the costs for the tests used in the base-case model. The EAG commented that costs of testing were reducing over time. It had used cost estimates from a micro-costing study (Ryan et al. 2019) in a scenario analysis to investigate this, which were much lower than the base-case values. Using these lower costs improved the cost effectiveness of testing for Lynch syndrome. The committee concluded that there was some uncertainty about the true costs associated with testing for Lynch syndrome.

4.11 The EAG did not find any evidence on the effect on clinical outcomes of testing for Lynch syndrome when people were diagnosed with endometrial cancer. So, it used a linked-evidence approach to assess:

4.12 In the base case, all testing strategies had ICERs of under £20,000 per QALY gained compared with no testing. Excluding the scenario in which no benefit was assumed for colonoscopic surveillance on colorectal cancer incidence (which the committee considered was unrealistic, see section 4.6), almost all of the testing strategies had ICERs under £20,000 per QALY gained compared with no testing in scenario analyses. In the scenario in which no benefit for aspirin was assumed, some strategies had ICERs of just over £20,000 per QALY gained. The committee also recalled that the effect of cancer on health-related quality of life was likely to have been underestimated in the model (see section 4.9). If a greater effect of cancer was used this would have improved the cost effectiveness of the testing strategies. The committee concluded that the most likely ICER for testing for Lynch syndrome for people with endometrial cancer was likely to be less than £20,000 per QALY gained. So testing was likely to be a cost-effective use of NHS resources.

4.13 In its base case, the EAG used data from Lu et al. (2007) for the testing strategies' accuracy. This was because this study had individual patient data that could be used to estimate sensitivity and specificity for most of the strategies. But, because of a lack of data, the EAG had to make assumptions to estimate values for MSI testing then MLH1 promoter hypermethylation testing. The EAG explained that its assumption that 66% of MLH1 promoter hypermethylation tests done after MSI testing were correct was based on an estimate used in Snowsill et al. (2019) of the probability of MLH1 promoter hypermethylation in sporadic tumours with MSI. The committee noted that when test accuracy estimates from Lu et al. (2007), the PETALS study and a recent meta-analysis (Snowsill et al. 2019) were used in the model, IHC then MLH1 promoter hypermethylation testing was consistently the most cost-effective strategy. Accuracy estimates from Chao et al. (2019) had also been used in a scenario analysis (only for IHC alone and MSI alone). Unlike the other studies, sensitivity was higher for MSI than IHC in this study, and MSI testing and no testing extendedly dominated IHC testing. Clinical experts highlighted that the sensitivity estimate from this study was based on 6 people with Lynch syndrome (4 people whose Lynch syndrome was identified by IHC testing and 2 people whose Lynch syndrome was not identified). A clinical expert highlighted that the 2 people whose Lynch syndrome was not identified by IHC had mutations in either the MSH2 or MSH6 gene. They explained that pathogenic mutations in these genes in particular often show some expression on IHC, which can make identifying MMR deficiency more difficult. They highlighted that the sensitivity of IHC to detect such mutations depends on the expertise of the pathologist and will be improved by following guidance such as the British Association of Gynaecological Pathologist's guidance on interpretation of MMR IHC. The committee concluded that, based on the base case and scenario analyses, IHC then MLH1 promoter hypermethylation testing (then germline testing to confirm a Lynch syndrome diagnosis) was likely to be the most cost-effective strategy.

4.14 The committee considered under what circumstances MLH1 promoter hypermethylation testing should be done after IHC for MMR proteins. Clinical experts commented that MLH1 promoter hypermethylation testing may be done if there is loss of PMS2 expression alone. This could mean that people with Lynch syndrome caused by PMS2 mutations would be missed if detection of MLH1 promoter hypermethylation in the tumour prevents germline genetic testing that would identify PMS2 mutations. Clinical experts further commented that isolated loss of PMS2 protein expression is likely to be rare, and could be because of artefacts of MLH1 protein expression that are mistaken for normal expression. The committee concluded that there was too much uncertainty to recommend MLH1 promoter hypermethylation testing if IHC showed loss of PMS2 expression alone, and this should be done only if IHC also shows loss of MLH1 protein expression in the tumour.

4.15 Clinical experts noted that the recommended testing strategy would not detect people with Lynch syndrome caused by constitutional epimutation of MLH1 (that is, methylation of the MLH1 promoter that is present in all cells in the body and is not acquired in the tumour). The committee noted that the EAG had not included testing non-tumour tissue for MLH1 promoter hypermethylation, as well as tumour tissue, in its economic model because of a lack of data to inform the clinical effectiveness of this. Clinical experts commented that constitutional MLH1 promoter methylation is likely to be rare, although prevalence may be underestimated because of lack of testing to identify it. The committee concluded that based on the EAG's model it was not able to make a recommendation on testing for constitutional MLH1 promoter methylation.

4.16 Clinical experts emphasised the importance of quality assurance to ensure that IHC testing for MMR proteins is done correctly. They also highlighted the British Association of Gynaecological Pathologists' recommended terminology for reporting mismatch repair protein immunohistochemistry with or without MLH1 promoter methylation results. The committee concluded that laboratories doing IHC testing for MMR proteins, MLH1 promoter hypermethylation testing or germline genetic testing should take part in a recognised external quality assurance programme.

4.17 The committee noted that costs and other parameter estimates used in the economic model can change over time, which may affect the cost effectiveness of testing. For example, if more testing for Lynch syndrome increases the number of people with known Lynch syndrome in the general population, fewer people with the condition will need testing after endometrial cancer, and the prevalence of the condition for those tested will decrease. This could reduce the cost effectiveness of testing. However, the committee noted that in sensitivity analysis decreasing the prevalence of Lynch syndrome to 1.6% only increased the ICER for strategy 4 to about £13,500 per QALY gained. Clinical experts also highlighted ongoing research on further interventions that could be used for people with Lynch syndrome. The committee also noted that the costs of sequencing DNA are decreasing. It recognised that emerging developments, such as next generation sequencing to test for MSI as part of tumour characterisation, could improve sensitivity to detect MSI in endometrial tumours. The committee concluded that future developments may affect the clinical and cost effectiveness of testing strategies, or their relative cost effectiveness compared with each other. It also concluded that it is important to monitor future developments to identify if any changes to the recommendations are needed.

4.18 The committee considered it important to monitor the effect of adopting testing for Lynch syndrome for people with endometrial cancer. For example: