Genetic testing in motor neuron disease and frontotemporal dementia: a 5-year multicentre evaluation

Introduction Motor neuron disease (MND) and frontotemporal dementia (FTD) comprise a neurodegenerative disease spectrum. Genetic testing and counselling is complex in MND/FTD owing to incomplete penetrance, variable phenotype and variants of uncertain significance. Affected patients and unaffected relatives are commonly referred to clinical genetics to consider genetic testing. However, no consensus exists regarding how such genetic testing should best be undertaken and on which patients. Objective We sought to ascertain UK clinical genetics testing practice in MND/FTD referrals, with the aim of helping inform guideline development. Methods MND/FTD clinical genetics referrals comprising both affected patients and unaffected relatives between 2012 and 2016 were identified and a standardised proforma used to collate data from clinical records. Results 301 referrals (70 affected, 231 unaffected) were reviewed across 10 genetics centres. Previously identified familial variants were known in 107 cases and 58% subsequently underwent testing (8 of 8 diagnostic and 54 of 99 predictive). The median number of genetic counselling appointments was 2 for diagnostic and 4 for predictive testing. Importantly, application of current UK Genomic Test Directory eligibility criteria would not have resulted in detection of all pathogenic variants observed in this cohort. Conclusion We propose pragmatic MND/FTD genetic testing guidelines based on appropriate genetic counselling.


INTRODUCTION
Motor neuron disease (MND), also known as amyotrophic lateral sclerosis (ALS), is an incurable adultonset neurodegenerative disorder with a worldwide incidence of 1.75 per 100 000. 1 Some 5% of cases of MND appear familial, following autosomal dominant inheritance with variable penetrance, and pathogenic variants in over 30 different genes have been aetiologically implicated. 2 3 The most commonly involved genes in familial MND are C9orf72 (MIM: 614260; a hexanucleotide GGGGCC expansion in up to 40%), SOD1 (MIM: 147450; pathogenic variants in 12%), TARDBP (MIM: 605078; 4%) and FUS (MIM: 137070; 4%), with other genes each accounting for 1% or less. 4 Up to 23% of patients with MND also fulfil the diagnostic criteria for frontotemporal dementia (FTD), and up to half of patients with MND experience FTD-like symptoms and/or signs. 5 6 FTD and MND constitute a disease spectrum, and some 12.5% of patients with FTD are also diagnosed with MND, while up to 40% experience MND-like symptoms and/or signs. 7 FTD has an overall incidence of 2.7-4.1 per 100 000, and at least 10% of cases display clear autosomal dominant inheritance, while up to 40% exhibit some family history. 8 9 Pathogenic variants in at least 12 genes have been linked to FTD, the most frequent being the C9orf72 hexanucleotide expansion, present in up to 25%, and variants in GRN (MIM: 138945) and MAPT (MIM: 157140), accounting for up to 15% and 6% of familial cases, respectively. 10 With growing availability of genetic testing, it has become apparent that no well-defined guidelines exist as to which patients are most suitable for testing and which genes to test on a clinical diagnostic basis. Practice therefore varies between centres across both clinical genetics and neurology specialties, depending on local expertise and interests. With the introduction of the UK Genomic Medicine Service, which seeks to standardise testing provision across genetic disorders via its Genomic Test Directory, we sought to ascertain the MND/FTD genetic testing practice within UK clinical genetics services and guide the development of pragmatic genetic testing criteria.

METHODS
We conducted a service evaluation using a standardised proforma to collect data from patients referred to UK clinical genetics centres from 1 January 2012 to 31 December 2016. The dates chosen reflected a 5-year period coinciding with the introduction of genetic testing for the C9orf72 expansion. Clinical records relating to MND/FTD referrals were reviewed for referral source, disease status, family history, presence of known familial mutation, genetic counselling, genetic testing and results, and subsequent clinical referrals. Positive family history was taken as presence of at least two affected first-degree or second-degree relatives.

RESULTS
Over the 5-year period, 301 MND/FTD referrals with relevant data were received from across 10 UK clinical genetics centres: Wessex, Exeter, Birmingham, Wales, London (North East Thames), Bristol, Leeds, Nottingham, Oxford and Cambridge. General practice was the most common referral source (63%, 191 of 301), followed by neurology (24%, 71 of 301) and psychiatry (7%, 20 of 301). Details on referral source, disease status, familial variants, tests performed and appointment data are shown in figure 1, while overall patient numbers and outcomes for this evaluation are shown in figure 2.
Affected patients comprised 23% of referrals (70 of 301). Of these, 35 had FTD (50%), 32 had MND (46%) and 3 had both diagnoses (4%). The majority of affected patients (67%, 47 of 70) also had a family history of MND/FTD. Of these individuals, eight (17%) already had a pathogenic variant known in their family. All eight subsequently underwent genetic testing and all were found to carry a pathogenic variant (seven cases of C9orf72 expansions and one SOD1 variant). Interestingly, of the 39 remaining affected individuals with a family history but no previously known familial variant, 11 (28%) did not undergo genetic testing, while pathogenic variants were subsequently identified in 46% (13 of 28) of individuals who did have testing. These comprised 11 C9orf72 expansions and 2 SOD1 variants. One-third of affected patients (23 of 70) had no family history of MND/FTD and no known familial pathogenic variant. Of these affected individuals, seven (30%) did not have genetic testing. Diagnostic testing was undertaken in the other 16 and a pathogenic C9orf72 expansion was identified in 4 cases (25%).
Unaffected relatives of affected individuals comprised 77% of referrals (231 of 301). Of these, 99 (43%) had a previously known pathogenic variant in the family. Just over half of these individuals (55%, 54 of 99) chose to undergo predictive genetic testing, while 45% (45 of 99) did not. Out of 54 unaffected relatives who had predictive testing, a pathogenic variant was identified in 25 (46%): 23 C9orf72 expansions (including 2 of intermediate size), 1 SOD1 variant and 1 MAPT variant. Notably, in the absence of a known familial pathogenic variant for which to predictively test, no unaffected relatives underwent blind genetic testing, irrespective of whether or not there was a positive family history of two or more affected individuals. Interestingly, 51% of unaffected relative referrals (117 of 231) only had one affected relative in the family, although 40 of these (34%) had a known familial pathogenic variant.
Overall, 36% (107 of 301) of referred patients had a pathogenic variant already known in their family. This was most commonly the C9orf72 expansion (76%, 81 of 107), while SOD1 comprised 6% (6 of 107) and MAPT and GRN each comprised 5% (5 of 107). Additional familial pathogenic variants in this cohort were in TARDBP (three variants), FUS and VCP (MIM: 601023; one variant each). C9orf72 was the most frequently requested diagnostic genetic test (96%, 50 of 52). Testing for a panel of genes (either linked to dementia/MND) was the next most common (31%, 16 of 52), followed by combined GRN and MAPT testing (10%, 5 of 52) and SOD1 testing (8%, 4 of 52). As expected, C9orf72 was the most frequent predictive test undertaken in unaffected relatives (87%, 47 of 54). SOD1 and TARDBP were each 4% (2 of 54) of the total predictive tests, Genetic counselling appointment data were available from 8 of 10 centres. Diagnostic test patients generally had one to two genetic counselling appointments (mean 1.8), while predictive test patients tended to have three to four appointments (mean 3.9) inclusive of results appointments and any follow-up provided. Affected patients were also more likely to have seen neurology/psychiatry prior to clinical genetics (89%, 62 of 70), compared with unaffected relatives (12%, 28 of 231). Comparatively, only 9% (20 of 231) of unaffected relatives were referred on to neurology; this includes those who underwent predictive testing and were found to be carriers and also some individuals with neurological symptoms who did not undergo testing.

DISCUSSION
Genetic testing is undergoing rapid change owing to advances in sequencing technology. Within the UK, access to testing across England is now subject to the criteria listed within the National Health Service Genomic Test Directory. 11 Relevant criteria for MND and FTD come under section 'R58 Adult onset neurodegenerative disorder'. On this basis, genetic testing in MND requires an ALS diagnosis with evidence of upper and lower motor neuron involvement, a progressive course, and additionally either age of onset under 40 years or an MND/FTD family history. Given the mean age of onset for MND/FTD in C9orf72 carriers is around 58 years, 40 years is likely too stringent a cut-off and risks denying diagnostic genetic testing to expansion carriers, especially as reduced and age-related penetrance frequently obscure dominant inheritance patterns. 12 13 Similarly, current criteria for FTD (or otherwise unexplained dementia) require age of onset under 55 years or else family history of the same dementia type in a first-degree or second-degree relative. This again likely represents too strict an age limit and additionally does not allow for MND family history. Although our evaluation did not collect age of onset data, based on Genomic Test Directory family history criteria alone, only 10 of 22 diagnostically tested individuals with FTD (45%), 13 of 19 with MND (68%) and 2 of 3 with MND and FTD (67%) in our cohort would have been eligible for testing, while 20 of 44 (45%) diagnostically tested individuals would not have been tested on account of 16 having no family history and 4 having family history not meeting the criteria. Furthermore, 5 cases (4 of 13 with FTD and 1 of 3 with MND and FTD) in this cohort would not have had their pathogenic C9orf72 expansions detected, suggesting the current criteria may only have a diagnostic sensitivity of around 71% (33% in FTD).
The genetics of MND and FTD are complex and evolving and the instigation of strict testing eligibility criteria as currently proposed may be premature. The paramount concern of any guidelines should be the best care and management of patients. Patients are of course individuals with their own specific circumstances and best practice requires that clinicians have flexibility to be able to treat them as such. This individuality is illustrated by the proportions of affected and unaffected individuals opting for/against testing and the diverse number of genetic counselling appointments observed. The role of clinicians in these circumstances is to help patients make the best choice for themselves and their families regarding genetic testing. The testing of sporadic MND or FTD cases presents a particular challenge with respect to family implications, and the need to consider testing in such patients is increasingly being recognised in neurology settings. 14 15 In this regard, collaboration between clinical genetics and neurology is highly recommended and would best be provided via multidisciplinary clinics, whereby neurology could provide detailed phenotyping of affected patients and unaffected (but potentially oligosymptomatic) relatives and also advise on disease-specific management, while genetics could contribute expertise in pretest and post-test counselling, variant interpretation, reproductive options and wider family management. A further consideration is the development of novel therapies for MND/FTD,

Neurogenetics
with clinical trials of antisense oligonucleotides for C9orf72 and SOD1 currently underway. Patients are increasingly aware of these research developments, which can influence their desire to undergo testing. Chiò et al 14 proposed an MND genetic testing algorithm involving two dimensions of clinical relevance and availability of effective treatments. If new genetic therapies do become available, this will significantly impact the uptake of presymptomatic genetic testing. In summary, we propose an approach to genetic testing in MND and FTD that is flexible enough to take account of patientspecific differences and which relies on adequate genetic counselling.

Familial pathogenic variant known
Offering targeted genetic testing on a diagnostic or predictive basis would be appropriate following adequate genetic counselling. For predictive testing this would usually involve a process akin to Huntington disease predictive testing. 16

Affected individuals with autosomal dominant inheritance (allowing for incomplete penetrance)
Sequential testing of C9orf72 followed by broader gene panelbased testing would be indicated following appropriate counselling. Such counselling must discuss not only personal and familial implications of the result but also the possibility of incomplete penetrance, the phenotypic spectrum of disease and the potential for and implications of finding variants of uncertain significance (VUS).

Affected individual without family history
Diagnostic testing of C9orf72 might be considered, but a cautious approach is advisable and adequate genetic counselling essential. While C9orf72 expansions occur in up to 7% of sporadic MND and up to 6% of sporadic FTD cases, population studies also suggest the expansion is present in 0.2%-0.6% of North Europeans, a frequency some 10 times greater than expected based on MND/FTD incidence, therefore indicating reduced penetrance. 17 18 Sequencing other genes in sporadic cases remains problematic owing to difficulty interpreting VUS and is therefore better avoided unless specific phenotypes suggest a genetic aetiology. In sporadic MND, SOD1 pathogenic variants occur in 1%-2% of cases, TARDBP and FUS pathogenic variants in around 1% each, with other genes implicated more rarely. 4 In sporadic FTD, GRN pathogenic variants occur in up to 5% of cases, MAPT in up to 3% and other genes more rarely. 19 20 There therefore remains a significant chance of finding pathogenic variants should testing be undertaken. However, in sporadic cases the positive predictive value of subsequently testing for such variants in unaffected relatives is likely reduced, since additional unknown protective factors may exist in a family that could contribute to reduced penetrance. Such calculations depend greatly on individual family structure and are best considered on a case-by-case basis.

Unaffected relatives without known familial pathogenic variant
Predictive genetic testing would not be considered appropriate given our limited understanding of penetrance, genetic heterogeneity and potential oligogenic effects.