Lynch syndrome is defined by the identification of constitutional pathogenic variants in four mismatch repair (MMR) genes (MLH1, MSH2, MSH6, PMS2). Although functional MMR deficiency (dMMR) is a hallmark of cancers associated with Lynch syndrome, dMMR may also be a feature of sporadic cancers as a consequence of somatic aberrations impacting MMR gene function.1 dMMR, as evidenced by abnormal expression of one or more MMR proteins, and/or microsatellite instability (MSI) is most commonly detected in colorectal (12%–15%) and endometrial (20%–30%) cancers but can be a feature of multiple other cancer types, including upper gastrointestinal2 (~10%), ovarian (~10%), hepatobiliary (10%) and adrenocortical cancers.3 4 Given that dMMR cancers typically demonstrate favourable response to immune checkpoint inhibitors (ICIs),5 the expanding availability of such therapeutic agents has fuelled increasing testing for, and detection of, dMMR in a host of cancer types, including common and rare Lynch syndrome-associated malignancies, as well as cancers not usually considered part of the Lynch syndrome phenotypic spectrum.6

The majority of dMMR detected in colorectal and endometrial cancers is attributable to sporadic somatic hypermethylation of the MLH1 promoter,7 8 and, usually, detection of hypermethylation of the MLH1 promoter in tumour-derived DNA (or, in colorectal cancer, detection of a pathogenic somatic BRAF variant as a surrogate marker of same) will obviate the need for constitutional genetic testing.9 Testing for constitutional hypermethylation of the MLH1 promoter, if not undertaken at the time of initial tumour-based analysis, and/or constitutional MMR gene testing may still be warranted if clinical suspicion for Lynch syndrome, based on patient age and/or family history, persists.10

At present, in the UK, constitutional and tumour-based genetic testing is undertaken, for the most part, in an unsynchronised, unpaired manner, driven, in part, by increasing awareness of Lynch syndrome as an underlying aetiology for a small but significant proportion of cancer risk but also by increasing application of and access to ICIs. The National Institute for Health and Care Excellence (NICE) has published guidance for assessment of MMR in endometrial and colorectal cancers, with investigative pathway guided by tumour type and, where available, pattern of MMR protein loss (figure 1).11 12 Detection of a pathogenic activating BRAF variant in colorectal cancer demonstrating loss of MLH1 (usually with concomitant loss of PMS2) can be used as a surrogate marker of MLH1 promoter hypermethylation, but 18%–20% of cancers demonstrating MLH1 promoter hypermethylation will not harbour a detectable BRAF variant.9 13 BRAF variants are rare in endometrial cancers, such that BRAF testing is not usually informative for such cancers.14 The utility of such testing in other extracolonic cancers is uncertain, as is the contribution of MLH1 promoter hypermethylation to dMMR in other tumour types.

Pathway for investigation of mismatch repair (MMR) and/or microsatellite instability in colorectal cancers. Decisions about the extent of investigation along this pathway are guided by clinical suspicion and the availability of relevant investigations. If investigation is stopped following germline MMR gene testing, the proportion of unexplained MMR deficiency will be higher.

For patients with cancers demonstrating combined MLH1/PMS2 deficiency and/or microsatellite instability where MLH1 promoter hypermethylation is not detected, or for patients with cancers demonstrating isolated loss of PMS2 or MSH6, or combined loss of MSH2 and MSH6, or where clinical suspicion for Lynch syndrome persists based on age and/or family history, constitutional MMR gene testing is indicated.11 12 Type and extent of constitutional testing should be guided by the pattern of protein loss and the a priori suspicion that a constitutional pathogenic variant explains the patient phenotype and in line with criteria outlined in the NHS England Genomic Testing Directory15 or equivalent in Wales, Northern Ireland and Scotland.16 In patients with tumours demonstrating isolated PMS2 deficiency, testing should include long-range PCR of PMS2 (or equivalent assessment of region impacted by pseudogene interference), while for tumours demonstrating loss of MSH2 and/or MSH6, testing should include dosage analysis of 3′ end of EPCAM, and testing for the recurrent paracentric inversion of exons 1–7 in MSH2.17 Where constitutional testing is uninformative, testing on tumour-derived DNA (‘somatic testing’) may then be undertaken to determine if an underlying somatic aetiology accounts for dMMR.18 In those cases where biallelic somatic events (biallelic variants or a single variant with evidence of loss of heterozygosity) impacting an MMR gene are not identified, the dMMR is classified as ‘unexplained’ (henceforth abbreviated as u-dMMR), and, historically, affected individuals were said to have ‘Lynch-like syndrome’ (LLS). The proportion of cases deemed ‘unexplained’ is influenced by the extent of investigation—after somatic BRAF sequencing and MLH1 promoter hypermethylation and, where indicated, constitutional MMR gene testing is undertaken, approximately 4.24% of colorectal cancers have u-dMMR; when somatic MMR analysis is undertaken, this reduces to 0.61%.19

Possible explanations for u-dMMR are undetectable somatic or constitutional pathogenic variants.20 Other explanations may include inaccuracies in the tests performed. For example, the performance of immunohistochemistry (IHC) is dependent on platform (including antibodies) and interpretation of results is observer-dependent and may be influenced by tumour fixation. Therefore, repeating an apparently abnormal IHC result may be useful as a first step in cases where an aetiology cannot be identified, particularly if IHC results are discordant with MSI results, and if clinical suspicion of Lynch syndrome is low.21 22 Testing an alternative well-fixed specimen, for example, diagnostic biopsy may be helpful to avoid the reduction/loss of staining noted with poorly fixed resection specimens. Consideration should also be given to the possibility of BRAF variants and/or MLH1 hypermethylation being missed by testing a sample with inadequate tumour cellularity.23

Published recommendations for bowel screening in relatives of affected probands deemed to have LLS are equivalent to that offered to patients with Lynch syndrome.24 However, while a proportion may be due to an unidentified heritable event, where clinical suspicion of a hereditary predisposition is low, it is more likely that dMMR will be due to an unidentified somatic aetiology.25 26 This is evidenced by the fact that relatives of probands with LLS have a lower cancer risk than in those families with molecularly confirmed Lynch syndrome, although higher than in those families with sporadic colorectal cancer.27

In a publicly funded health system, it could be argued that it is most appropriate that limited resources are best focused on detection of rare, atypical mechanisms of heritable dMMR20 and screening of relatives of probands where suspicion of hereditable aetiology remains high after uninformative investigations. However, conversely, as screening recommendations can be ‘downgraded’ once an aetiology for MMR is established, it could be argued that further investigations could be warranted where an explanation for MMR deficiency is not immediately obvious but where clinical suspicion of constitutional mechanism is low.

Every two months, a virtual multidisciplinary meeting involving clinical geneticists, genetic counsellors and other healthcare professionals working in Cancer Genetics convenes as part of a national collaborative initiative between the Cancer Research UK-funded CanGene-CanVar Programme and the UK Cancer Genetics Group (UKCGG), a constituent group of the British Society of Genomic Medicine.28 Each meeting has a specific theme, and clinicians are invited to submit challenging cases aligned with that theme. The session in November 2022 was on the theme of u-dMMR. A number of cases were presented (table 1), all of whom were patients with dMMR cancers where aetiology was not established. The audience was asked to comment on bowel screening recommendations based on clinical features (eg, age of cancer diagnoses, cancer type, family history and results of tumour and constitutional testing). Recommendations ranged from population bowel cancer screening for relatives to Lynch-equivalent bowel screening.

Cases discussed at UKCGG/CanGene-CanVar Cancer Genetics MDT

During the meeting, consensus was easily achieved for bowel screening recommendations for each case, although this was only informally recorded given the nature of the meeting. From the discussions, it was noted that most centres consider tumour type, age and family history when determining extent of testing to establish aetiology of dMMR and screening recommendations for relatives. However, it was noted that regional variability in access to certain technologies such as long-range PCR for PMS2 and loss of heterozygosity analysis was a limitation.

In cases where constitutional MMR testing is uninformative and clinical suspicion of Lynch syndrome low, there was discussion about whether the cumulative human resources and cost required in undertaking additional tumour-based testing to try to determine aetiology for dMMR is unacceptably high when considering diagnostic yield. It was questioned whether this would be cost-effective considering numbers needed to test/screen to prevent one cancer-related death. It was suggested that tumour-based testing to establish somatic aetiology of dMMR might be most appropriate in those cases where clinical suspicion of a heritable predisposition is at least moderate, in order to determine the most appropriate familial bowel screening recommendations; and ensure limited resources and highest intensity screening are directed to those families where suspicion for Lynch syndrome remains high. To more fully address these issues, the UKCGG convened a national consensus meeting on 25 April 2023.

Key experts representing gastroenterology, pathology, gynaecology and clinical genetics from each genomic medicine service alliance and each of the four countries in the UK were invited to contribute to a national consensus meeting. Participants from the Republic of Ireland were also permitted to attend. In advance of the meeting, an electronic survey was circulated to cancer genetics leads in each regional clinical genetics service, to establish current practices across the UK.

Specifically, attendees were asked to comment on:

Their approach to different clinical scenarios regarding colorectal surveillance strategies

The extent to which testing was completed according to the algorithm in figure 1, according to patients of different ages with typical versus atypical Lynch syndrome spectrum tumours (online supplemental table 1)

Access to testing in their centres (online supplemental figure 1), and what testing they felt should be available.

The factors that should be considered in determining the likelihood of a heritable cause of dMMR cancer, for example, age and tumour type.

When molecular testing could be considered completed in order to make clinical recommendations (ie, was ‘exhaustive’ testing required as standard or could a more pragmatic approach be taken?)

Key expert speakers were invited to give an overview of current evidence on prevalence of, and approaches to testing for, dMMR across a host of tumour types. The agenda of the meeting is outlined in appendix 1. These talks provided the multidisciplinary audience with relevant overview of evidence on which to base their responses to proposed statements. Recordings of these talks and copies of speakers’ slides are available to download (https://www.ukcgg.org/information-education/ukcgg-consensus-meetings/). In-meeting polls were then run to achieve consensus on a number of statements generated based on thematic analysis of survey results. In line with standard Delphi processes, consensus was deemed to be achieved when at least 50 responses (approximately half of respondents to which relevant statements applied) were recorded, and when at least 80% of responses were ‘agree/strongly agree’ with the proposed statement.29 Open discussion was encouraged where minimum number of responses or consensus was not achieved, or if audience members indicated strong feelings in contravention of the proposed statement. This approach has been successfully implemented for several other UKCGG-led national consensus meetings to generate clinical guidelines.30–34

The term LLS refers to an intermediate state of testing, and not a distinct heritable syndrome per se, and has therefore never been clearly defined despite advances in molecular testing and evaluation. Thus, some authors have suggested that the term may cause confusion and is redundant and should not be used.35 37 38 The consensus in this meeting was that the term ‘u-dMMR’ more accurately reflected intermediate and/or uncertainty in testing for dMMR cancers.

There remain significant diagnostic and management challenges when considering patients with cancers demonstrating u-dMMR. Although it was agreed by participants that the term LLS was ill-defined and obsolete, this term has been used in clinical guidelines for colorectal cancer screening in relatives of probands with u-dMMR cancers. The redefinition of LLS to ‘u- dMMR’ by this consensus meeting therefore necessitates a reassessment of existing clinical guidelines24 to align with our consensus indicating that Lynch syndrome-equivalent surveillance is not always appropriate or necessary in families where one individual is affected by u-dMMR cancer.

Significant discussion was generated regarding the extent of testing required to ‘rule out’ Lynch syndrome before surveillance recommendations could be made, with participants citing issues related to access to certain technologies and resources required to perform such testing. Many participants also highlighted the different paradigms in arranging tests to ‘exclude’ a hereditary predisposition, compared with most clinical scenarios where resources are generally focused on diagnosis of heritable mechanisms of disease. The likelihood of Lynch syndrome can be significantly reduced in the majority of patients with dMMR cancers following constitutional testing, and therefore a pragmatic approach to additional testing and ongoing surveillance for family members may be justifiable in some situations. Limitations of genomic testing not unique to Lynch syndrome but to all inherited conditions are such that heritable risk factors can never be fully excluded, and the considerable resources required to investigate dMMR cancers over and above an uninformative germline test, where the prior probability of Lynch syndrome is low must be considered.

Much discussion and variance in opinion focused on lack of data regarding proportion of non-colorectal/non-endometrial dMMR cancers accounted for by Lynch syndrome. This was clearly an issue when considering statements related to extent of investigation/screening recommendations for probands with non-colorectal/non-endometrial cancers.

Although reconfiguration of Genomic Medicine Services in NHS England in 2019 was intended to standardise availability and application of genomic testing in different clinical scenarios across England,39 it is clear that variability in both access and use of different technologies exists between genomic laboratory hubs in England, not to mention between different nations within the UK. Lack of routine access to loss of heterozygosity analysis was flagged as one particular issue hampering comprehensive investigation of probands presenting with dMMR, such that individuals with dMMR tumours demonstrating a single somatic variant and uninformative germline genetic testing are managed in the same way as those individuals where no causative variants are identified in constitutional or tumour-derived DNA. This is likely inappropriate as a significant proportion of tumours in which a single relevant variant will also demonstrate loss of heterozygosity at the locus of interest. Inequitable access to, and application of, long-range PCR testing of PMS2 was also identified as an issue, although consensus was reached that this should be implemented as part of standard testing for those individuals with tumours demonstrating isolated PMS2 loss.

In 2023, the National Lynch Syndrome Registry was successfully implemented in England, where entry to the register requires a molecular confirmation of Lynch syndrome.40 We did not discuss how we can prospectively follow probands with u-dMMR, but it is clear that such data are required to inform guidelines in the future. Furthermore, data are required to determine proportion of dMMR in non-colorectal/non-endometrial cancers accounted for by various somatic aetiologies, not to mention clinical and analytical sensitivity and specificity of available assays to detect underlying mechanisms of disease.

Data so far indicate that the number of patients with dMMR colorectal cancer appropriately referred for germline genetic testing is suboptimal,40 41 and this issue may be even more pronounced for patients with those cancers less commonly recognised as a Lynch syndrome-associated malignancy. Further training and education are required to ensure timely diagnosis of Lynch syndrome, and, where resources are limited, this may represent a more acute clinical risk than scenarios related to u-dMMR where significant genomic testing has already been undertaken.

As assessment of MMR function continues to expand across different tumour types and constitutional testing of MMR genes increases in patients with non-colorectal/non-endometrial cancers, we hope to generate robust data regarding proportion of dMMR in such cancers accounted for by Lynch syndrome or alternative genomic mechanisms, by continuing and building on previous collaborative work between NHS genomic laboratory hubs and National Disease Registration Service.42 We appreciate, however, that a lack of uniformity in investigation of u-dMMR where clinical suspicion is low will hamper data collection for such cases—further emphasising the need for national collaboration to enable collection of at least some meaningful data. This information about the outcomes of testing, especially for non-colorectal/non-endometrial cancers, will help to inform diagnostic strategies, for example, clarifying the diagnostic yield of somatic and/or germline testing in these tumours.

Limitations of time at this meeting precluded discussion of gynaecological cancer risk management in probands with dMMR cancers and their at-risk relatives. Furthermore, we did not consider indications for consideration of constitutional mosaicism.

While it is not possible to reach uniform advice for all u-dMMR cases, the authors hope that this article describes the difficulties and limitations encountered in clinical practice and provides the reader with support in clinical decision making for individual cases. Recommendations regarding extent of investigation and colorectal cancer surveillance for probands with dMMR cancers and their at-risk relatives should be made based on the best available evidence; but it is crucial that collection of evidence should be practicable in the face of limited resources (money, staff and time).

Despite efforts in standardising availability of comprehensive testing, inequity of access to the same is evident across the UK, hampering efforts to develop uniform guidance for best practice. As a compromise, if it is not reasonably feasible to perform comprehensive constitutional and tumour-based testing for every patient with dMMR cancer, bearing in mind local resources, identification of heritable risk factors in those patients where Mendelian cancer predisposition is most suspected should be prioritised. Consideration of further testing for cryptic constitutional MMR gene variants (currently only available on a research basis) should be considered where clinical suspicion of Lynch syndrome persists—for example, with those patients fulfilling Amsterdam II43/Revised Bethesda44 criteria, or based on a high PREMM (PREdiction Model for gene Mutations)36 score.20 In such cases, Lynch-equivalent colonoscopic screening may be appropriate for FDRs of affected individuals.

Clinical judgement should be used to inform extent to which testing beyond standard testing is performed in families where clinical suspicion of such predisposition is lower. Importantly, clinical judgement should be used to inform surveillance recommendations in probands and their family members, and Lynch-equivalent surveillance is not always indicated where aetiology of dMMR remains elusive after reasonable efforts to determine the same.

We recommend that, at a minimum, ‘reasonable effort’ should include (where relevant), somatic BRAF testing and/or MLH1 promoter hypermethylation testing and constitutional MMR gene testing (and additional genes if indicated as per national genomic testing directories), with further testing beyond that depending on clinical circumstances as outlined above.

Basic principles should also be considered, and it is important that performance and reporting of results from IHC and/or genomic testing be undertaken in line with best practice.45–47