Introduction

One of the latest developments in amyotrophic lateral sclerosis (ALS) is the emergence of genotype-specific pharmacotherapies heralding a paradigm shift from generic neuroprotective strategies to precision, genotype-specific interventions [1, 2]. The fundamental heterogeneity of the disease is now universally recognised and genotype- and phenotype-specific clinical traits, radiological signatures and disease trajectories have been characterised [3,4,5]. The notion that a long presymptomatic phase precedes symptom manifestation and that neurodevelopmental factors may contribute to the pathogenesis of ALS is increasingly accepted [6]. The presymptomatic phase of ALS has been relatively arcane until the publication of seminal presymptomatic papers which confirmed considerable pathological changes years before symptom manifestation [7, 8]. Existing presymptomatic studies in ALS invariably suffer from sample size limitations and are strikingly diverse with regards to their study design, methodological approach and conclusions. The systematic review of these papers, the frank discussion of their limitations, and the careful integration of their findings is particularly timely with the emergence of genotype-specific therapies [1, 2]. One of the key contributions of recent imaging studies in ALS is the confirmation that by the time patients fulfil diagnostic criteria for ALS, considerable disease burden can already be detected, limiting the therapeutic potential of putative disease-modifying drugs. These observations suggest that the optimal therapeutic window is likely to be at an earlier stage in high risk, genetically susceptible populations. The presymptomatic phase of ALS has long been of academic interest [9], and inspired small-scale studies [10], dedicated terminology [9], but the advent of antisense oligonucleotide therapies (ASOs) highlights the urgency for large presymptomatic studies and the meticulous integration of molecular, pathological and radiological observations in asymptomatic mutation carriers. ASO-mediated drugs have already been approved by the US Food and Drug Administration for the treatment of spinal muscular atrophy and Duchenne muscular dystrophy [11,12,13], and currently being trialled for ALS [1]. The nuanced characterisation of pathophysiological processes before perceptible disability develops may help to identify the optimal therapeutic window for pharmacological intervention before irreversible functional impairment ensues. Longitudinal studies of mutation carriers spanning from the adolescence to significant disability would provide an opportunity to describe anatomical patterns of disease spread, validate current staging systems, evaluate prognostic indicators and test prevailing propagation theories such as corticofugal spread, network-wise propagation, selective vulnerability, trans-synaptic spread etc., [9, 14,15,16]. Reports of considerable presymptomatic cerebral pathology without overt functional impairment also suggest a degree of ‘motor reserve’, network redundancy or possible compensatory processes to maintain function until a critical threshold is reached and symptoms develop. Large presymptomatic studies also permit the systematic assessment of the sensitivity profile of our current biomarkers. For example, detecting white matter changes in a patient with significant disability may be less challenging than capturing early asymptomatic changes in white matter integrity decade before projected symptom manifestation. Despite the dual academic and clinical relevance of characterising the presymptomatic course of ALS [17,18,19], striking inconsistencies exist in the current literature due to the sample size limitations and methodological differences [20,21,22]. Our objective is the careful integration of the lessons of existing presymptomatic studies in ALS, to identify key learning points from individual studies, reflect on common methodological shortcomings and distil a robust framework for future studies.

Methods

A formal systematic literature review was conducted using the following search terms on PubMed: “amyotrophic lateral sclerosis”, “motor neuron disease”, “C9orf72”, “frontotemporal lobar degeneration”, “frontotemporal dementia” combined with each of the following keywords “presymptomatic”, “premanifest”, “asymptomatic”. An additional search combined the above search terms with the following keywords: “magnetic resonance imaging”, “MRI”, “positron emission tomography”, “PET”, “electromyography”, “neuroimaging”, “electrophysiology”, “neurophysiology”, “transcranial magnetic stimulation”, “motor unit number estimation”, “motor unit number index”, “neurofilament”, “biomarkers”. Only original research papers were systematically reviewed. Conference abstracts published in supplements of neuroscience journals were not considered. Only human studies were systematically reviewed. No exclusion criteria were set based on year of publication, but only articles written in English were reviewed. Animal studies, review papers, opinion pieces, editorials, case reports, and case series were excluded. Based on the above criteria a total of 48 original research papers were reviewed in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) recommendations. The identified papers were systematically reviewed based on the following criteria:

  1. (1)

    Core study design parameters (target cohort: pre-symptomatic/pre-manifest, control group: healthy controls/non-carrier relatives/disease controls, sample size, cross sectional/longitudinal, prospective/retrospective, multi-centre/single centre, follow-up interval, number of follow-up time points, number of participants, attrition rates, follow-up into the symptomatic phase)

  2. (2)

    Clinical and laboratory assessments (Genetic testing, demographic profiles, availability of electrophysiological assessment, neurological or neuropsychological data, functional rating scales)

  3. (3)

    Neuroimaging methods (cerebral/spinal, whole-brain/region-of-interest, MRI / PET, structural/diffusion/spectroscopy/rs-fMRI, field strength 1,5 T/3 T / 7 T, single/multi-modal analyses, post-mortem validation.

To appraise methodological approaches in other neurodegenerative conditions and their potential applicability to ALS, a selection of presymptomatic imaging papers were also reviewed in frontotemporal dementia (20 papers), Huntington’s disease (11 papers), Alzheimer’s disease (13 papers) and Parkinson’s disease (13 papers).

Results

While the terms ‘preclinical’, ‘premanifest’ and ‘presymptomatic’ are widely used, the term ‘asymptomatic mutation carrier’ is preferred by many. It has been proposed [20] that term ‘preclinical’ should be reserved for the period where there are no identifiable pathological changes and the term ‘presymptomatic’ used for the phase when neuroimaging, electrophysiology or detailed cognitive assessment may already detect abnormalities. Despite these recommendations the above terms are often used interchangeably. The number of papers identified stratified by the key study methodology are shown in Fig. 1.

Fig. 1
figure 1

Number of studies identified by the core search criteria

Full size image

Neuroimaging studies of asymptomatic mutation carriers

Cohort characteristics

Based on our search criteria, we have identified twenty-five imaging studies investigating presymptomatic C9orf72 hexanucleotide carriers [7, 8, 23,24,25,26,27,28,29,30], four studies focusing on presymptomatic SOD1 carriers [10, 30,31,32,33,34,35], and three studies evaluating both [30,31,32]. We also identified a study which included NEK1, TARDBP and FUS gene mutation carriers in addition to participants with the C9orf72 and the SOD1 [31]. The number of presymptomatic subjects included in single studies showed significant variation ranging from 2 [35] to 249 [36]. Eight studies had included over 100 presymptomatic carriers [36,37,38,39,40,41,42]. The number of presymptomatic C9orf72 carriers included ranges from 2 [32] to 83 [28] and the number of presymptomatic SOD1 carriers ranges from 2 [35] to 24 [10] in the current literature. Thirteen studies also included symptomatic patients [28, 40, 43,44,45], and 19 studies only focussed their investigation on presymptomatic cohorts [8, 27, 29, 37, 38]. The strategy to select symptomatic participants was inconsistent; some studies only included symptomatic gene carriers [10, 23, 24, 28, 31, 35, 40, 41, 43,44,45,46] while others included sporadic patients [24, 32]. The size of the symptomatic cohort also shows great variation ranging from as little as 12 subjects [32] to 270 [31]. Symptomatic control cohorts included patients with ALS [10, 24, 28, 31, 35], FTD [24, 28] and ALS-FTD [24, 28]. Without exception, all identified studies included a cohort of healthy controls. These cohorts were either unrelated healthy controls [10, 24, 28, 29, 31,32,33, 35, 40], or more commonly, gene negative relatives of symptomatic patients [8, 23, 34, 36, 37, 42, 47]. A common shortcoming of the available papers is that when unrelated healthy controls were used as a reference group, their gene profile is seldom reported [30]. Furthermore, none of the reviewed studies used ‘disease controls’, which would have helped to gauge the specificity of findings to ALS. The mean age of asymptomatic C9orf72 subjects ranged from 39.8 [7] to 51 years [29] and in the case of SOD1, carriers from 32.3 [34] to 47.2 years [10, 30]. In studies where a symptomatic cohort was included, their mean age ranged was 47.8 years [32] to 65.2 years [41]. Presymptomatic cohorts were generally relatives of symptomatic ALS [8, 27, 32], ALS-FTD or FTD patients [8, 27, 36] (Table 1).

Table 1 A selection of presymptomatic neuroimaging studies in ALS/FTD
Full size table

Cohort size observations

Existing presymptomatic studies vary considerably with regards to overall sample size and statistical power. The total number of participants ranges from as few as 21 [34] to as many as 472 subjects [36]. While several studies included over 300 participants, these are invariably multi-centre studies necessitating some degree of inter-rater reliability testing for clinical assessments, sequence harmonisation for imaging and standard operating procedures for biomarker collection, storage and analysis [28, 31, 36, 40, 41]. Irrespective of the overall sample size, almost half all identified presymptomatic studies in ALS-FTD (15 out of 32) resulted from data generated from multi-site consortia such as the Genetic Frontotemporal dementia Initiative (GENFI) [28, 36, 38,39,40,41,42,43,44,45,46], the ARTFL–LEFFTDS Longitudinal Frontotemporal Lobar Degeneration (ALLFTD) [37, 47] research consortium and the Predict to Prevent Frontotemporal Lobar Degeneration and Amyotrophic Lateral Sclerosis Study Group (PREV-DEMALS) [7]. More than half of the identified studies (17 out of 32) were single-centre studies [10, 23,24,25,26, 32,33,34,35, 48]. Very few single-centre studies reached cohort sizes over 100 [49,50,51], and the largest single-centre recruited included 113 participants [50]. The most asymptomatic C9orf72 participants included in a single-centre study were 40 participants [8] (Table 1).

Study methods

The vast majority, 22 out of the 32 presymptomatic imaging studies are cross-sectional [10, 33, 39, 45, 48], and only ten longitudinal presymptomatic studies can be presently identified [8, 24, 28, 30, 38, 43, 44, 47, 50, 51]. Significant variability can also be observed with regards to follow-up intervals, which range from 6 months [24] to 2 years [50, 51]. The longest overall follow-up period was 6 years [51]. The majority of longitudinal studies are 2 time-point studies, with a select few assessing subjects across three timepoints [24, 47, 51]. Only one study evaluated participants up to four times longitudinally [28]. T1-weighted structural MRI data were appraised in the majority (24) of presymptomatic imaging studies [28, 37, 38, 40,41,42,43, 45,46,47], diffusion MRI data in 14 studies [7, 8, 23, 27, 31, 33, 34] and functional MRI data in 6 studies [25, 32, 36, 44, 49, 51]. MR spectroscopy was performed in one study [10] and one study used arterial spin labelling [39]. Two presymptomatic PET studies were identified [29, 35], one using flumazenil [35], and the other used fluorodeoxyglucose (F-FDG) [29] as a tracer. While many studies investigated a single parameter, 14 studies implemented a multi-modal approach [7, 23,24,25,26, 29, 32,33,34, 39, 49,50,51]. The majority of MRI studies were performed on a 3 T MRI platform [7, 8, 10, 28, 33]. Seven studies relied on imaging data from a 1.5 T scanner [34] and of these, 5 were multi-site studies relying on mixed data from 1.5 T and 3 T scanners [31, 38, 40, 41, 44]. No ultra-high field (7 T) human presymptomatic studies were identified at the time of this review (Table 1).

Imaging findings in presymptomatic ALS

In C9orf72 hexanucleotide repeat expansion carriers, frontal [28], temporal, parietal, occipital [23], thalamic [25, 26], cerebellar [26] and striatal [25] atrophy was consistently detected. Diffusion MRI captured reduced WM integrity in the corticospinal tracts [8, 25], orbitofrontal regions [30], corpus callosum, cingulum, uncinate and inferior longitudinal fasciculi [7, 25, 27]. Neurite orientation dispersion and density imaging (NODDI) is thought to be more sensitive in detecting white matter alterations than traditional DTI metrics [27]. A PET study of presymptomatic C9orf72 participants confirmed hypometabolism in frontotemporal, insular, thalamic and basal ganglia regions [29]. It is noteworthy, that some studies did not detect cerebral changes in asymptomatic C9orf72 carriers [24].

In asymptomatic SOD1 carriers, white matter degeneration was observed in the posterior limb of the internal capsule [34], reduced flumazenil binding in the left fronto-temporal junction [35], and reduced NAA/Cr and NAA/Myo ratios in the superior spinal cord [10]. A multimodal study of seven asymptomatic SOD1 carriers found no significant abnormalities on diffusion tensor imaging and threshold tracking transcranial magnetic stimulation [33].

In asymptomatic mixed-genotype cohorts frontal, temporal, parietal [48] and cerebellar [41] atrophy was noted as well as subcortical grey matter degeneration including the caudate [48], hypothalamus [31], and thalamus [41, 48]. White matter alterations were observed in the anterior thalamic radiation [50]. Marked connectivity changes were detected by some functional MRI studies [32, 36], while others identified functional resilience despite structural degeneration [44] (Table 1).

The clinical profile of mutation carriers

Accompanying clinical assessments

Most presymptomatic imaging studies incorporate a brief neurological assessment to screen for clinical signs [24, 27, 29, 31, 32, 34, 35, 37, 49,50,51], but the details of the exam are seldom reported [8, 10, 25, 48]. In symptomatic patients, the revised amyotrophic lateral sclerosis functional rating scale (ALSFRS-r) [8, 10, 24, 31,32,33, 35], Medical Research Council (MRC) scale [33], Trigg’s hand function score [33], and composite upper motor neuron (UMN) scores [35] are typically administered. Nineteen out of 32 studies also commented on neuropsychological performance. The battery of neuropsychological instruments varied greatly across the identified studies. Many studies used generic, non-ALS specific, screening tests such as the Mini Mental State Examination (MMSE) [7, 8, 24, 25, 27,28,29, 36, 38, 41, 42, 48,49,50,51], the Montreal Cognitive Assessment (MOCA) [47], the Frontal Assessment Battery (FAB) [7, 27, 48], the Clinical Dementia Rating Scale (CDR) [47], or the Mattis Dementia rating scale (MDRS). Some centres relied on ALS-specific screening tools such as the (ECAS) [25, 29, 30], while others used an extensive battery of neuropsychological tests assessing memory, visuospatial, language, executive domains, anxiety and depression [25]. Some studies focused on specific cognitive domains known to be preferentially affected in ALS, such as executive function. This was typically interrogated by digit span [42], Stroop test [26], trail making test (TMT) [36, 42, 47], fluency tasks [7, 26, 42], the Delis–Kaplan Executive Function System (D-KEFS) [24] or symbol digit modalities test (SDMT) [42]. Language was either assessed by the Boston naming test [7, 42] or the Wide Range Achievement Test (WRAT) [25]. Memory performance was appraised using the California Verbal Learning test (CVLT) and Benson figure recall [25] and other recall tests [7, 42]. Visuospatial function was assessed using the Benson figure and The Visual Object and Space Perception Battery (VOSP) [25]. Possible neuropsychiatric manifestations have been assessed by the Neuropsychiatric Inventory Questionnaire (NPI-Q) [25, 50], and depression has been screened for by the geriatric depression scale [25] or the Beck Depression Inventory (BDI) [25, 29]. While deficits in social cognition are recognised in symptomatic ALS, these are seldom assessed specifically in presymptomatic cohorts [30, 53,54,55]. Presymptomatic behavioural manifestations were evaluated by the revised Cambridge Behavioural Inventory (CBI-R), [28, 36, 42] and the Frontal Behavioural Inventory (FBI) [7, 24, 48]. ALS-specific behavioural instruments [56, 57] were not applied to presymptomatic cohorts. Other instruments used in presymptomatic studies included the neuropsychological battery of the Uniform Data Set (UDSNB) and the Executive Abilities: Measures and Instruments for Neurobehavioral Evaluation and Research (EXAMINER) [47].

Clinical findings in presymptomatic cohorts

A study of presymptomatic C9orf72 carriers identified subtle deficits in executive functioning, verbal fluency [30] and memory [29], while another study detected significant memory impairment [25]. Several studies found that MMSE scores are slightly lower in presymptomatic C9orf72 carriers, but still within normal range [7, 25, 28, 29, 48]. Performance on other tests such as the MDRS [7], FAB [7, 48], FBI [7, 48], Benson figure [7, 25], Boston naming test [7, 25], Stroop, verbal fluency, and digit span [25] is thought to be relatively preserved and comparable to healthy controls.

Electrophysiology studies of asymptomatic mutation carriers

Presymptomatic mutation carriers at risk of developing ALS have also been extensively investigated by quantitative electrophysiology tools (Table 2), such as motor unit number estimation (MUNE) [21, 22, 58], magnetoencephalography (MEG) [59], and transcranial magnetic stimulation (TMS) [33, 60,61,62]. Some of these studies also interrogated accompanying MRI data [33, 59, 62]. Asymptomatic C9orf72 [59, 60, 62] and SOD1 [21, 22, 33, 58, 59, 61] mutation carriers are typically either compared to unrelated healthy controls [33, 59,60,61], gene-negative family members [62] or both [21, 22, 58]. Longitudinal studies [21, 60, 62] followed patients for up to 3 years [21, 60] with a follow-up interval as short as 6 months [21]. The number of total participants in presymptomatic electrophysiology studies range from 52 [59] to 186 [62] with up to 19 presymptomatic SOD1 [21, 58] and up to 11 presymptomatic C9orf72 [60] carriers included in any one study. The majority of these studies also include a group of symptomatic patients [21, 22, 59,60,61], and often symptomatic mutation carriers [62]. Symptomatic patients were mostly ALS [21, 22, 59,60,61] or FTD patients [62], but one study also included PLS patients [59]. None of the reviewed studies included neurodegenerative ‘disease-controls’. The age profile of presymptomatic mutation carriers in electrophysiology studies range from 40 [61] to 51.7 [59] but several studies did not report demographic data in detail. In presymptomatic electrophysiology studies, the presymptomatic cohort was on average 10 years younger than the symptomatic cohort. On TMS, alterations in intracortical facilitation transmission are seen up to 3 decades before expected symptom onset [62]. One study showed that SICI was absent in only two presymptomatic SOD1 carriers and reduced in one [61]. A study investigating MUNE showed no changes in presymptomatic SOD1 carriers [22], but a follow-up study reported that 2 of 19 SOD1 carriers showed reduction in MUNE just months before the symptom onset [21]. Cortical hyperexcitability was detected in symptomatic C9orf72 carriers but not asymptomatic carriers [60]. Some electrophysiology studies report the clinical profile, including MRC scores in presymptomatic cohorts [21, 58, 61] or ALSFRS-r in symptomatic patients [59,60,61]. Cognitive screening with ECAS [59] or MMSE [62] was also implemented in some studies.

Table 2 A selection of presymptomatic electrophysiology studies in ALS
Full size table

Biofluid studies of asymptomatic mutation carriers

In the era of ‘omics’ (proteomics, lipidomics, metabolomics etc.) biofluid markers are also increasingly evaluated in asymptomatic mutation carriers including neurofilament light (NfL) [63,64,65] or (NEFL) [66], neurofilament heavy (pNfH) [64, 65, 67] or NEFH [66], poly(GP) proteins [68, 69], chitotriosidase-1 (CHIT1) [66, 67, 70], chitinase 3-like protein 1 (CHI3L1) [66, 67], chitinase 3-like protein 2 (CHI3L2) [67], C-reactive protein (CRP) [67], mitochondrial DNA (mtDNA) [71], ubiquitin carboxyl-terminal hydrolase 1 (UCHL1) [66, 70], microtubule-associated protein 2 (MAP2) [66], macrophage-capping protein (CAPG) [66], glycoprotein non-metastatic B (GPNMB) [66], histone cluster 1, H4 (HIST1H4A) [66], histone cluster 1, H2b (HIST1H2B) [66], neurofilament medium (NEFM) [66, 70], neuronal pentraxin receptor (NPTXR) [70]. Some studies focused on protein profiles in a single biofluid either in the CSF [66, 67, 69, 70] or serum [71], while others evalauted both CSF and serum [63,64,65, 68]. The most commonly used methods to quantify the concentration of these markers are enzyme-linked immunosorbent assays (ELISA) [64, 65, 67], electrochemiluminescence immunoassay (ECLIA) [63, 65], mass spectroscopy [66, 70], meso scale discovery-based immunoassay [68] and poly-GP immunoassay [69]. Some longitudinal studies, followed asymptomatic mutation carriers for over 3 years [63, 64], and in some studies controls were also assessed longitudinally [63, 64]. Most biofluid studies investigated asymptomatic SOD1 [63,64,65,66, 71] and C9orf72 [63,64,65,66,67,68,69,70,71] cohorts, but some included TARDBP [65, 66] or FUS [65] mutation carriers. The biggest biofluid study included 84 subjects, including 52 SOD1 and 27 C9orf72 hexanucleotide carriers among other mutation carriers [63]. All identified studies also concurrently assessed a symptomatic cohort of either ALS [63,64,65,66,67,68,69,70], ALS-FTD [68] or FTD [69] patients. Some studies included a PLS cohort [67, 68], as well as disease controls such as Alzheimer’s disease [68, 69], Lewy body dementia [68], Parkinson’s disease [69], or Kennedy’s disease [67]. Mean age of asymptomatic mutation carriers ranged from 39.7 [67] to 48.3 years [65]. Some biofluid studies screened for the presence of UMN and LMN signs [65]. In symptomatic patients, ALSFRS-r was invariably recorded, and some studies also reported UMN [64, 67], ECAS [64, 67] or FTD-CDR scores [69]. Presymptomatic neurofilament studies are inconsistent; many did not detect elevated levels [65,66,67]. Altered protein profiles were identified by some studies [68,69,70], especially in the years preceding phenoconversion [63, 64] (Table 3).

Table 3 A selection of presymptomatic biofluid studies
Full size table

Lessons from other neurodegenerative conditions

Despite the clinical differences, frontotemporal dementia studies offer ample learning opportunities for ALS study designs. In addition C9orf72, other FTD-associated mutations have been extensively investigated, including GRN [72,73,74,75,76,77,78,79,80,81,82,83,84], MAPT [75, 85,86,87,88] and CHMP2B [89,90,91] carriers. Mutation carriers were more commonly compared to gene-negative first-degree relatives [72, 75, 76, 78,79,80, 88,89,90,91], but also to healthy controls [73, 74, 77, 81, 86, 87] or both [82, 84]. The average age of presymptomatic cohorts in FTD ranges from 31 [88] to 56 years [91]. Many of the reviewed studies also investigated a symptomatic cohort, but a few only focussed on their presymptomatic cohort [74,75,76, 84, 87, 89,90,91]. Similarly to ALS, no studies included ‘disease controls’. Neuropsychological data was more commonly included than in ALS studies, including screening tests such as the CDR [72, 79,80,81], MOCA [72] and MMSE [73, 74, 78, 82, 87], behavioural tools such as the FBI [78] and FAB [78], neuropsychiatric instruments such as the NPI [78, 82] or BDI [87], as well as executive [73,74,75, 82, 87], language [73, 74, 87], visuospatial [73, 82] and memory tests [73, 82, 87, 88]. While a minority of studies investigated a single imaging parameter [72, 80, 89], most studies presented structural data as well as diffusion [74, 75, 84, 86], functional [73, 75, 77, 79, 81, 82] or PET data [76, 78, 85, 91]. Most of the longitudinal studies were two time-point studies [72, 74, 76, 87, 89,90,91], but one multi-timepoint study followed patients over 11 years [88]. Robust multi-centre initiatives such as Longitudinal Evaluation of Familial Frontotemporal Dementia Subjects (LEFFTDS) [72, 88] and The Genetic Frontotemporal dementia Initiative (GENFI) have been particularly successful at gathering large datasets. One of the most widely studied neurodegenerative condition in its presymptomatic phase is Huntington’s disease which has been extensively evaluated by structural [92,93,94,95,96,97,98], diffusion [95, 99] and functional [97, 100] MRI studies. These studies tend to be much larger than ALS studies and often include data from hundreds of participants [92, 94]. Multi-centre initiatives such as PREDICT-HD [92, 94], IMAGE-HD [93, 98] and TRACK-On HD [95, 97] have facilitated these robust collaborative studies. In presymptomatic Alzheimer’s studies, APP [101,102,103,104,105], PSEN1 [101,102,103,104,105,106,107], PSEN2 [101,102,103,104,105] and APOE4 [101, 104, 105, 108,109,110] mutation carriers were investigated with some studies including over 300 participants [102, 103]. In presymptomatic Parkinson’s disease, LRRK2 [111,112,113,114,115,116,117,118], parkin [114, 118,119,120], PINK1 [118], ATP13A2 [118] and SNCA [121] mutation carriers were evaluated. These studies are relatively smaller than those conducted in HD and AD, but can include as much as 130 participants [118].

Discussion

Irrespective of their methodology and genetic focus, the majority of existing presymptomatic ALS studies have confirmed considerable biological changes before symptom manifestation. They are also consistent in identifying changes in brain regions which are characteristically affected in symptomatic mutation carriers. The demographic analysis of existing C9orf72 studies highlights that mutation carriers in their 30 s already exhibit considerable structural degeneration, decades before typical symptom manifestation [7, 8]. The considerable pathological changes detected in young mutation carriers raises questions about neurodevelopmental factors [6, 30, 122], but this could only be appraised if mutation carriers in their teens and twenties were also included subject to appropriate approvals and genetic counselling. Existing studies suggest a relatively divergent imaging signature in asymptomatic C9orf72 and SOD1 carriers, but in the absence of well-powered studies including large cohorts of both mutations, these genotype-specific traits are not firmly established. Nonetheless, the presymptomatic signature of C9orf72 seems to be associated with more widespread frontotemporal and subcortical grey matter degeneration than those observed in association with SOD1 [7, 27]. While the available studies are conceptually important, stereotypical shortcomings can be readily identified. The sample size of asymptomatic mutation carriers in ALS imaging studies range from 2 to 83, which coupled with the considerable variation in the age of participants, prevents making conclusive observations regarding presymptomatic biology. The sample size limitations of presymptomatic ALS studies are particularly striking in contrast to the available AD and HD literature. The systematic review of the literature also highlights that in contrast to presymptomatic AD and HD studies, the majority of presymptomatic studies in ALS are single-centre, or national studies. Several ALS studies have included both SOD1 and C9orf72 carriers in a single presymptomatic group to demonstrate structural changes before symptom manifestation. However, with the emergence of ASO therapies, it seems paramount to describe genotype-specific changes in a specific presymptomatic mutation cohort rather than characterising admixed cohorts. The importance of defining genetically homogenous groups cannot be underestimated, as carriers of specific mutation may exhibit different rate of progression and anatomical involvement. The strategy to ascertain presymptomatic changes also varies considerably in ALS studies; some research groups contrast their presymptomatic cohort to healthy controls alone, others to symptomatic patients, and others to gene-negative family members. Another common problem is the selection of symptomatic patients. Including a symptomatic patient cohort in presymptomatic studies is particularly useful if they carry the same mutation as the presymptomatic cohort, but the inclusion of a mixed gene-positive and gene-negative sporadic patients, or symptomatic patients without genetic profiling hinders data interpretability. If the symptomatic cohort carries the same mutation, their “affected” brain regions can be specifically evaluated in the presymptomatic group in targeted region-of-interest analyses. Another potential shortcoming of existing studies is the lack of disease controls which makes it difficult to appraise how specific the findings are to ALS or to a given genotype. For example, corpus callosum degeneration is regarded as pathognomonic of ALS by the ALS research community, despite being observed in a range of other conditions such as HSP to AD. Similarly, increased neurofilament levels were observed in a number of presymptomatic ALS studies, but they are also raised in many other neurological conditions. If no disease-controls are included, the specificity of a biomarker to ALS is impossible to assess and its ability to distinguish between neurodegenerative processes remain questionable. For example, if a proposed biomarker such as CSF neurofilaments, corticospinal tract FA or a TMS parameter is similarly affected in ALS, PLS and CBD, it will not distinguish between these conditions rendering their diagnostic value relatively limited. In real-life clinical scenarios, the question is seldom whether a patient is healthy or not, the question is typically whether the constellation of symptoms presage ALS or PLS or CBD, as these conditions carry distinctly different prognoses. Not only existing studies do not include disease controls, they do not include other motor neuron disease phenotypes either. This seems like a missed opportunity. The inclusion of relatively pure UMN and LMN phenotypes, such as adult SMA or PLS may help to gauge the sensitivity of a proposed marker to the UMN versus LMN system [123,124,125]. Another contentious aspect of existing studies is the generalisation of observations from small presymptomatic C9orf72 and SOD1 cohorts as representative of presymptomatic ALS as a whole. Depending on the population, the vast majority of ALS patients test negative for ALS-associated mutations and are seemingly sporadic. Accordingly, the presymptomatic phase of sporadic patients remains a conundrum and may differ in anatomical involvement and chronological dynamics from the traits observed in C9orf72 or SOD1. This also applies to PLS, which is not closely linked to single mutations and very little is known about cerebral and spinal disease burden prior to symptom onset. PLS exhibits overlapping albeit UMN predominant clinical and radiological characteristics with ALS [123], and in the absence of specific mutations to carry out presymptomatic studies, research groups attempted to characterise ‘early’ symptomatic cohorts before fulfilling diagnostic criteria [124, 126]. These therefore cannot be regarded as presymptomatic studies, rather pre-diagnosis studies of suspected cohorts. The lessons of these studies can be integrated in future ALS studies, namely that gene negative ‘suspected ALS’ patients should also be included in imaging and biomarker studies in an attempt to characterise early pathology in sporadic cohorts. Additionally, presymptomatic gene-positive cohorts should be followed beyond symptom manifestation and at least until they fulfil current diagnostic criteria for ALS. Characterising disease burden by quantitative imaging, electrophysiology and wet biomarker protocols at the time of fulfilling diagnostic criteria, may help to highlight the limitations of existing diagnostic criteria. The refinement of diagnostic criteria may enable an earlier diagnosis in suspected patients and in turn earlier inclusion in clinical trials. One aspiration would be the introduction of ‘radiologically-supported ALS’ based on objective radiological variables. Another important question is the optimal timing ASO therapy. Once brain and cord pathology is detected and electrophysiology changes ascertained in mutation carriers there may be an argument to introduce therapy early before widespread irreversible changes ensue. Existing presymptomatic studies also raise important question re: motor reserve. The observation that both electrophysiology and MRI detects considerable pyramidal tract, motor cortex and spinal cord degeneration long before projected symptom manifestation suggest a degree of network resilience or redundancy. A simplistic interpretation may be that compensatory processes and redundant networks offset degenerative changes until a critical threshold is reached.

While cross-sectional studies have provided pioneering insights, they provide limited information on progressive changes as they average radiological signatures across different age groups. With few exceptions [7], existing presymptomatic studies in ALS rely on convenience samples of asymptomatic mutation carriers with considerable dispersion in their demographic profile. For the meaningful analysis of early alterations, demographically homogenous samples would be desirable. Longitudinal studies have the potential to provide a nuanced picture of progressive changes; map anatomical propagation patterns, progressive functional alterations and evolving CSF/serum signatures overtime. However these studies should also ideally recruit demographically homogenous cohorts. The limitations of two timepoint longitudinal designs are also clear, as these don’t permit the modelling of non-linear changes and the assessment of ceiling- and flooring-effects [127]. The lessons of large AD and HD studies also apply here, namely that large multi-timepoint designs are necessary to characterise progressive structural degeneration in ALS. A limitation of existing longitudinal presymptomatic studies in ALS is that mutation carriers are seldom followed until phenoconversion or beyond. The availability of radiological, electrophysiology or wet biomarker panel in mutation carriers before and after symptom manifestation would also permit the evaluation of prognostic indicators. There are two practical deliverables which were not addressed by existing studies, both of which seem relevant for individualised patient care and future pharmacological trials. One of them is the estimation of projected age of symptom onset based on disease burden in the presymptomatic phase. This would be possible if mutation carriers would be meticulously followed until symptom manifestation. The other practical aspect of presymptomatic studies pertains to C9orf72 from a phenotype point-of-view; namely can patterns of cerebral or spinal cord involvement be used to predict if an individual GGGGCC repeat expansion carrier is more likely to develop FTD or ALS (ALS-FTD). Clinical and radiological data have been previously used to build prognostic models for individual symptomatic patients, but these are yet to be applied to presymptomatic individuals [128,129,130,131]. These observations highlight another shortcoming of existing presymptomatic studies in ALS; with very few exceptions [8, 10] nearly all presymptomatic radiology studies are brain studies. Spinal cord involvement is a key aspect of ALS, which encompasses anterior horn (LMN) and descending pyramidal tract (UMN) degeneration and is now readily detected by novel imaging applications [132, 133]. Given the availability of robust quantitative spinal protocols, these should be carefully integrated into future presymptomatic studies to assess if they detect changes earlier than brain protocols and if they can presage age of onset, site of onset, or UMN/LMN predominance. It is conceivable that a hexanucleotide carrier with ample extra-motor cerebral involvement with no spinal cord abnormalities is more likely to develop FTD, than ALS, but unless such studies are conducted and patients followed until disease manifestations the predictive value of presymptomatic imaging is difficult to gauge. The practical deliverables of robust presymptomatic studies therefore include phenotypic prediction, age of onset estimation and optimising the timing of pharmacological interventions Table 4. Very few presymptomatic studies report negative or unexpected findings [33, 60]. The candid reporting of negative results is hugely important as they either reveal genotype-specific traits or reflect on the detection sensitivity of the methods implemented. Similarly, the comparative evaluation of several imaging metrics in the same cohort is helpful to appraise the detection sensitivity of specific methods. Some pioneering spectroscopy studies for example did not perform accompanying structural assessments and vice versa [10]. With the current imaging technology at our disposal the detection of white and grey matter alterations in symptomatic ALS cohorts is no longer challenging [134, 135], but the concomitant implementation of several imaging modalities in the presymptomatic phase enables the critical comparison of various techniques. For example NODDI is thought to be superior to characterise white matter degeneration in presymptomatic cohorts than standard DTI [27]. Multimodal longitudinal imaging in presymptomatic cohorts may additionally help to establish if certain imaging indices exhibit early ceiling effect [127] which would limit its utility to track the post-symptomatic changes in clinical trials [136]. Robust presymptomatic studies can also deliver on important academic objectives. A myriad of environmental factors have been proposed in ALS which could be objectively evaluated in vivo if mutation carriers were tracked from a young age over multiple timepoints and environmental factors would carefully recorded.

Table 4 Lessons of existing presymptomatic studies in ALS, learning points for future study designs
Full size table

Conclusions

From an academic perspective, presymptomatic studies offer invaluable learning opportunities to study propagation patterns, characterise early genotype-associated signatures, assess functional resilience, explore concepts like “motor reserve” or “cognitive reserve”, and evaluate neurodevelopmental or environmental factors. However, with the advent of ASO therapies, the meticulous study of presymptomatic cohorts in ALS gained practical relevance and unprecedented urgency. Future studies have to be designed to address specific clinical objectives such as informing the timing of pharmacological interventions, monitoring response to therapy, validating phenotypic indicators, and develop novel, biomarker-supported diagnostic criteria to facilitate earlier entry in clinical trials.