Classifying ALS: towards a more logical future
Ammar Al-Chalabi PhD FRCP1*, Orla Hardiman MD FRCPI2, Matthew C Kiernan PhD FRACP3, Adriano Chio MD4, Benjamin Rix-Brooks MD5, Leonard van den Berg MD PhD6
1. Maurice Wohl Clinical Neuroscience Institute, Department of Basic and Clinical Neuroscience, King’s College London, London SE5 9NU, UK
2. Academic Unit of Neurology, Biomedical Sciences Institute, Trinity College Dublin, Ireland 3. Brain and Mind Centre; and Sydney Medical School, The University of Sydney, Sydney, Australia 4. “Rita Levi Montalcini” Department of Neuroscience, University of Turin, Turin, Italy
5. Carolinas Neuromuscular/ALS-MDA Center, Department of Neurology, Carolinas Medical Center, Carolinas Healthcare System Neurosciences Institute, and the University of North Carolina School of Medicine - Charlotte Campus, Charlotte, NC 28207-1885, USA
6. Department of Neurology, Brain Center Rudolf Magnus, University Medical Center Utrecht, Netherlands
*Address for Correspondence: Ammar Al-Chalabi, Maurice Wohl Clinical Neuroscience Institute, King’s College London, London SE5 9NU, UK
Email: ammar.al-chalabi@kcl.ac.uk Telephone: +44 20 7848 5174 Fax: +44 20 7848 5190
Summary
Amyotrophic lateral sclerosis is an incurable progressive adult onset neurodegenerative disease, primarily affecting upper and lower motor neurons. There is considerable variability in onset, progression and clinical pattern, so that in the early stages, the diagnosis may not be obvious. Furthermore, there is no definitive test, yet it is a devastating diagnosis to give. Clinical descriptions are therefore used both to make the diagnosis and to classify the different clinical patterns. This dual use of clinical descriptions has resulted in overlap between diagnostic labels and phenotypic labels, resulting in mixed or inconsistent use of some terms. The problem is compounded by the fact that most clinical features show continuous variation while the labels are categories without a clear boundary between one category and the next. A logical classification of ALS with separation between diagnosis and phenotypic classification is important for patient communication, optimising
Introduction
The description of amyotrophic lateral sclerosis (ALS) as “a progressive neurological disease in which upper and lower motor neurons degenerate, leading to relentlessly worsening paralysis of voluntary muscles until death ensues”,[A: ref? **]
(http://www.ninds.nih.gov/disorders/amyotrophiclateralsclerosis/detail_ALS.htm) is a definition that most neurologists would recognise. Familiarity with the disease however quickly reveals that there is great clinical variability in presentation and prognosis. This variability may mean that a logical, consistent description of clinically defined subtypes is not possible. However, for several reasons [A: please could you briefly state these reasons if not covered in the following text? **], generating a classification of ALS is important, even though it is not straightforward.{Rutter-Locher, 2016 #354} Classification helps guide treatment, corresponds to prognosis, allows clinical trials and other research to analyse more homogeneous groups for a more personalised approach to therapy, and is valued by patients and their families. [A: it would be helpful if you could mention in the following text how diagnosis is currently made – ie, based on clinical (and EMG?) findings?**] There is no definitive test for ALS, and confirmation of diagnosis is based on the clinical picture, EMG findings, and the exclusion of mimics. There is a potential inconsistency when subtypes are defined [A: please could you mention the different subtypes here to provide readers with the key terms needed to follow the rest of the paper (see also notes on history below)?**], since they can be arbitrarily assigned as a new disease or as an extension of the existing ALS spectrum. The upper motor neuron disease primary lateral sclerosis (PLS) and lower motor neuron disease progressive muscular atrophy (PMA) were initially regarded as separate entities, {Aran, 1850 #71;Cruveilhier, 1853 #72} but were subsequently considered different manifestations of the same condition {Brain, 1933 #73} described by the term “motor neuron disease”.{Brain, 1962 #74} This cohorting of three different clinical phenotypes into a unifying diagnosis, based on degeneration of different
components of the motor system, has been used by clinicians for over 50 years. However, a key difficulty remains as to whether PLS and PMA are distinct diseases or part of the spectrum of ALS, {Mackay, 1963 #340} and therefore whether “ALS” really means degenerative motor neuron diseases as a whole.{Hu, 1998 #56;Shoesmith, 2007 #65;Wijesekera, 2009 #68;Sabatelli, 2008 #63;Turner, 2013 #301} The situation has been complicated by the use of the terms bulbar palsy, pseudobulbar palsy and progressive bulbar palsy to describe anatomically circumscribed patterns of ALS, PLS and PMA confined to musculature controlling speaking and swallowing, and terms such as flail arm or flail leg syndrome to describe anatomically distinct patterns, since these descriptions mix anatomy, neural level, and clinical presentation.
The problem is further compounded by the rapid increase in knowledge of genetic causes, a deeper understanding of the non-motor manifestations of ALS, and the discovery of pathological subtypes that overlap partly but not neatly with the genetic and clinical patterns [A: refs?**].{Al-Chalabi, 2012 #150} There is also imprecision in the terms motor neuron disease more generally and ALS in particular, terms that to some extent are now used interchangeably but also as diagnostic or
phenotypic descriptors. [A: it would be helpful if you could introduce the current criteria to provide context for the subsequent discussion and, briefly, the need for revised criteria/classification systems**] Currently, there is no universal system for ALS classification. The method with the greatest consensus is classification by the El Escorial criteria, which use the degree and distribution of upper and lower motor neuron signs, but the criteria are limited in scope and not suitable for all applications.
In this Personal View, we will summarize the subtypes of ALS based on different approaches to classification, and highlight the challenges raised by advances in our understanding of the disease process itself. [A: please could you say more about your goals for this paper and your vision for the Personal View, including a brief mention of your proposed classification? Ideally, the introduction should entice the readers in, build up the rationale for this Personal View, and, ideally, include a
timely angle (eg, mention ongoing revision of current classification/diagnostic systems?)**] We show that existing classifications are inconsistent, illogical and unable to successfully describe ALS or its phenotypes, as evidenced for example, by the repeated revisions of the El Escorial criteria or the International Classification of Diseases section on motor neuron diseases. We use the latest research findings to propose a more logical approach, able to impart relevant information while being simple to use, familiar enough to ALS practitioners to be taken on easily, yet remaining flexible enough to incorporate new research findings as they come about.
Outline of the criteria for a classification system [A: please see my email about this section] There are two basic requirements of an ALS classification system. The first is related to diagnosis. What are the criteria for ALS as opposed to other disorders? The second is related to the underlying subtypes of ALS. How should we think about the different features of ALS in relation to classification? Both of these questions present problems. As is the case for all rare conditions, the challenge is whether a diagnostic definition should be specific and potentially miss people who will later turn out to have ALS (false negative diagnosis), or be sensitive and potentially include people who will not (false positive diagnosis). This distinction has implications for patients because the diagnosis has such life changing consequences, and also affects access to health services. Definitive diagnosis is also a desired goal for patient recruitment into clinical trials, although more relaxed criteria could be justified in studies of compounds with minimal adverse effects, such as methylcobalamin [A: ref?**]. {Kaji, 2015 #355}
One must also consider the purpose of classification. From a research or academic viewpoint, the object of subclassification can be seen in the context of improving our understanding of disease heterogeneity, which might alter how we stratify for clinical trials, based on the evolving assumption that different subcohorts of patients with ALS might respond differently to treatments.
At a clinical level, ALS heterogeneity is characterized by variations in site of symptom onset (bulbar or spinal innervated muscles) and rate of progressive motor decline. However, it is now clear that these parameters are not sufficiently sensitive to fully reflect underlying differences in disease
pathophysiology or clinical outcome, leading to large variations within trial cohorts.{Beghi, 2007 #69} The heterogeneity of ALS could therefore be viewed as an impediment to new drug development, but extensive knowledge of disease heterogeneity should be harnessed to inform, rather than confound progress in understanding disease pathogenesis. This is the rationale behind a precision medicine based approach towards therapeutics in general, as subclassification is used to enhance effective drug discovery [A: ref?**].{Sciences, 2015 #356}
http://www.acmedsci.ac.uk/download.php?f=file&i=32644
History of the classification of ALS [A: please shorten this paragraph and include just a very brief history (a few sentences) in the introduction to outline the key terms and provide the background needed for readers to understand the subsequent text** reduced to a single sentence as the key terms are now in the introduction]
In the mid-19th century, it was understood that neurogenic weakness could result from damage to upper motor neurons or lower motor neurons, but it was not until 1869 that Charcot described ALS, based on neuropathological observation, as a combination of upper motor neuron disease (sclerosis of the lateral corticospinal tracts) with lower motor neuron disease and muscle wasting
(amyotrophy).{Charcot, 1869 #70}
Current diagnostic and classification systems The El Escorial criteria
Confirming a diagnosis of ALS remains the cornerstone around which the institution of appropriate management strategies can be instigated. These include commencement (and reimbursement) of
Riluzole, referral to multidisciplinary care programmes and provision of diagnostic certainty for patients and families to facilitate future planning.{Kiernan, 2011 #1;Vucic, 2014 #2} In the absence of a confirmatory clinical test for ALS, diagnosis has relied on research-based consensus criteria,
specifically the El Escorial criteria, developed by the ALS Research Group of the World Federation of Neurology to define clinical diagnostic criteria, initially at a meeting in El Escorial, Spain in 1990 [A: text added from below for context**], and the revisions, the Airlie House and Awaji criteria.{Brooks, 1994 #3;Brooks, 2000 #4;de Carvalho, 2008 #5} The criteria have varying levels of sensitivity and specificity,{de Carvalho, 2009 #11} limiting their usefulness as either diagnostic tools or staging devices.
As it currently stands, the diagnosis of ALS remains phenotypically based, relying on the clinical identification [A: is this done clinically?**] of upper and lower motor neuron signs within body regions defined as bulbar, cervical (corresponding to neck, arm, hand , diaphragm cervical spinal cord muscles), thoracic (corresponding to back and abdomen muscles), and lumbar (corresponding to back, abdomen, leg and foot lumbosacral spinal cord lumbosacral spinal cord innervated muscles), with the level of diagnostic certainty dependent on the extent of motor neuron dysfunction.{Brooks, 1994 #3;Brooks, 2000 #4;de Carvalho, 2008 #5} In the original criteria, there were four levels of diagnostic certainty ranging from Suspected ALS in which only lower motor neuron dysfunction was observed, to Definite ALS in which three body regions with mixed upper and lower motor neuron findings were observed [A: ref?].{Brooks, 1994 #3} The levels of diagnostic certainty related to the combined burden of disease as defined by the concatenation of clinical upper and lower motor signs, supplemented by neurophysiological [A: EMG?** it depends on the clinical picture, but usually yes] and imaging data [A: is this MRI and PET?** not usually PET. MRI, CT, etc but again depends on the clinical picture] to exclude ALS mimics.
In 1998, the El Escorial criteria were revised at a meeting at Airlie House to improve diagnostic sensitivity, by removing the Suspected category and adding a Laboratory Supported Probable ALS category, defined as the presence of upper and lower motor neuron signs in a region, with features of lower motor neuron dysfunction evident in two regions on neurophysiological testing.{Brooks, 2000 #4} Although the criteria were specific for ALS, sensitivity remained an issue, particularly in the early stages of the disease, resulting in significant diagnostic delay and limiting recruitment of ALS patients into therapeutic trials.{Aggarwal, 2008 #8;Chio, 1999 #6;Turner, 2009 #7}
In a further attempt to reduce diagnostic delay, neurophysiologically derived Awaji-Shima criteria were conceived in 2006.{de Carvalho, 2008 #5} These criteria proposed that neurophysiological features of lower motor neuron dysfunction, including acute changes such as fibrillation potentials, and chronic neurogenic changes such as unstable motor units, were equivalent to clinical features of lower motor neuron dysfunction. Separately, fasciculations were also identified as features of active denervation, with morphology used to define ALS-specific fasciculations. To date, identification of upper motor neuron dysfunction has remained clinically based, although progress in neuroimaging and functional approaches such as transcranial magnetic stimulation may significantly impact future strategies.
The diagnostic utility of the Awaji-Shima criteria has been compared to the Airlie House criteria through single centre studies, which have mostly identified an increased sensitivity when compared to the Airlie House criteria.{Boekestein, 2010 #12;Chen, 2010 #16;Costa, 2012 #14;de Carvalho, 2009 #11;Douglass, 2010 #10;Gawel, 2014 #18;Krarup, 2011 #17;Okita, 2011 #15;Schrooten, 2011 #13} The diagnostic benefits appear to be most helpful in bulbar-onset ALS.{Costa, 2012 #14;Jang, 2015 #19}
[A: could you mention in this paragraph that this system is also based on phenotypic classification to ensure that the logical flow is clear?**]The International Classification of Diseases (ICD) is a method to allow mortality and morbidity statistics to be compared across institutions and countries, and should reflect a logical description of each disease and its subtypes. The ICD is revised on a regular basis to reflect changes in medical understanding of disease. For ALS, the system is based on phenotypic classification. The original structure, provided by the World Federation of Neurology Research Group on Motor Neuron Diseases in the form of the El Escorial Criteria, is a core foundation that might advance the current revision of ICD-9 and ICD-10 (released in October 2015) [A: addition correct?**Yes] to a more specific linking of the diagnostic classifications in ICD-11 with standard terminologies such as the Systematized Nomenclature of Medicine-Clinical Terms (SNOMED-CT). {Rodrigues, 2013 #291;Rodrigues, 2015 #292;Kingdon, 2010 #293} Currently there is no clear ontology for defining the classification of motor neuron diseases in the transition from 9 to ICD-10, reflecting the lack of a logical structure to the classifications[A: please explain this point in the current context**]. [A: also mention classification according to DSM-5?** DSM-5 is for mental disorders so not very relevant for ALS]
Phenotypic classification: challenges and considerations [A: OK?**Yes]
[A: could you also mention challenges related to phenotypic classification with use of ICD (and DSM?) coding in this section intro?**] The problems with the use of ICD-9, ICD-10, El Escorial criteria and their revisions in diagnosing ALS remain those of correct identification of upper and lower motor neuron signs, especially when there is subclinical or minor involvement, and decisions around the inclusion of incomplete ALS phenotypes such as those with little in the way of upper motor neuron involvement. This is compounded by the use of phenotype for diagnosis when the range of phenotypes is broad, the underlying trait is a continuous variable, and there is clinical heterogeneity. [A: please could you relate the discussion of issues in the subsections below to the existing criteria/classification systems and critically assess the value of these criteria in these different contexts?**] Furthermore, the formal classification systems compete with the informal systems used by clinicians in every day practice or discussion with patients, which vary considerably between individuals.{Rutter-Locher, 2016 #354}
Clinical phenotypes based on motor involvement
The clinical phenotypes of ALS can be considered in two categories: complete forms of ALS with widespread involvement of upper and lower motor neurons, and incomplete forms of ALS in which the syndrome is confined to a subgroup of motor neurons. These incomplete forms seem to represent more than merely different clinical presentations, being characterized by impressive differences of gender ratio, age at onset and survival.{Chio, 2011 #81} In fact, while some
phenotypes have a similar distribution between genders (flail leg syndrome, upper motor neuron predominant ALS, PLS), others are significantly more frequent in men (ALS, PMA, flail arm syndrome, monomelic ALS and respiratory ALS).{Wijesekera, 2009 #68} The median age at onset is ten years lower in the incomplete forms than in all the other variants [A: ref?** In general I do not re-cite a citation that is already used in the same paragraph]. Yet, despite the clinical and prognostic
importance of these clinical patterns, they are not used at all in the El Escorial criteria, ***The PMA, PLS and UMN predominant variants, together with the flail arm variant, are also characterized by a less aggressive clinical course, with a significantly longer survival [A: ref? **same here – already cited (Wijesekera 2009)]. Reasons for the failure of the disease process to express clinically beyond these restricted regions are not well understood.
[A: I’ve moved this paragraph from the section on the El Escorial criteria and it would be helpful if you could incorporate these points alongside the relevant text in the discussion below**Done]
The utility of diagnostic criteria are limited in incomplete ALS phenotypes.{Ludolph, 2015 #20} Unlike lower motor neuron signs, which can be objectively detected by electromyography, upper motor neuron signs can be difficult to detect reliably using conventional neurophysiological measures or any neuroimaging approaches [A: which imaging approaches?***any], and there can be
considerable inter-observer variability in clinical measurement [A: ref?]. For example, a reflex that is objectively reduced in the context of severe muscle atrophy might be regarded as brisk by an experienced neurologist. Furthermore, upper motor neuron signs may develop late in the disease and an initial diagnosis of PMA may therefore change with time to ALS. Neuropathologic studies have supported Lord Brain’s classification of ALS, PMA, and PLS as a single disorder [A: addition correct? **Yes], as corticospinal degeneration can be demonstrated on autopsy in patients with PMA. {Tartaglia, 2007 #76;Turner, 2010 #66;Visser, 2007 #67} Notwithstanding, distinction is important because, for example, patients who have only lower motor neuron signs at a disease duration of at least 4 years may have a better prognosis.{van den Berg-Vos, 2003 #302} In addition, the most recent revision of the El Escorial criteria suggested that identification of lower motor neuron dysfunction in isolation may be deemed as being consistent with possible ALS, but requires attention to rule out ALS mimic disorders.{Ludolph, 2015 #20} Existing classification systems deal with incomplete forms of ALS rather crudely, for example, PMA is not considered within the El Escorial criteria at all.
For PLS there is a similar problem. Many patients presenting with pure upper motor neuron degeneration subsequently develop lower motor neuron signs,{Tartaglia, 2007 #76;Gordon, 2006 #77} but there appears to be a continuum, with some developing lower motor neuron degeneration that is only detectable on EMG, and others never seeming to develop lower motor neuron
involvement, although it is accepted that there remains a relative dearth of neuropathology data from PLS cohorts.{Short, 2005 #342} However, most patients (>75%) who initially present with only upper motor neuron signs eventually develop lower motor neuron signs within 4 years of symptom onset. For this reason, the diagnosis of PLS cannot be made with certainty at initial consultations.15 Prognosis appears to correlate with the degree of subsequent lower motor neuron degeneration, so that a pure upper motor neuron syndrome has the best prognosis, and those who go on to develop typical ALS have the worst.{Le Forestier, 2001 #78} Patients with predominantly upper motor neuron disease with minor lower motor neuron signs, so called upper motor neuron predominant ALS, have disability similar to ALS, but slower progression. PLS is not explicitly diagnosed within the El Escorial criteria, but merely grouped with “possible” ALS.
It is also important, but often difficult, to differentiate sporadic presentations of hereditary spastic paresis (HSP) from PLS without bulbar involvement. Differentiation can be based on clinical
characteristics, such as mild dorsal column impairment (decreased vibratory sense or abnormal leg somatosensory evoked potentials), symptoms of urinary urgency, or mild electromyographic abnormalities, but this is unreliable, and genetic testing for HSP related genes remains the most accurate method of separating these two phenotypically similar conditions.{Brugman, 2009 #303} Symmetry of the lower motor neuron phenotype appears to be important prognostically{Wijesekera, 2009 #68} and a classification has been proposed into PMA which is asymmetrical, and adult onset spinal muscular atrophy which is symmetrical and progresses much more slowly.{Visser, 2007 #67;Van den Berg-Vos, 2009 #79} Identification of subclinical upper motor neuron dysfunction with transcranial magnetic stimulation techniques has been shown to facilitate the diagnosis in the flail leg syndrome, a symmetrical predominantly lower motor neuron form of ALS affecting the legs and consistent with adult onset spinal muscular atrophy.{Vucic, 2007 #21;Swash, 2012 #85;Menon, 2015 #86;Vucic, 2011 #87} Features such as symmetry are not included in existing classifications, even though prognosis appears to depend on it.
Thus, existing methods of classification are focussed only on typical forms of ALS, but are inadequate when faced with phenotypic variations. Any new system should allow for such variation to be defined and when present, explicitly stated.
[A: for all these points, what are the implications for existing criteria/classification systems and for the development of new systems? Please provide evaluation of this throughout this section**] Bulbar palsy versus bulbar onset
Approximately 25% of people develop initial symptoms of ALS in the bulbar innervated muscles controlling speech and swallowing [A: ref?**].{ Chio, 2011 #81} If the weakness is lower motor neuron predominant, it is referred to as progressive bulbar palsy, and if upper motor neuron predominant, pseudobulbar palsy, with the whole referred to simply as bulbar palsy. These variants may progress to involve spinal innervated muscles but appear to have a worse prognosis than spinal forms.{Burrell, 2011 #304} A bulbar predominant phenotype has been associated with a higher frequency of cognitive impairment in some clinical series,{Hu, 2009 #57;Lomen-Hoerth, 2003 #58} although population-based studies are conflicting about this observation and it is likely an artefact of disease progression rates.{Phukan, 2012 #61;Montuschi, 2015 #60} Thus it makes sense on one level to consider bulbar palsy as a diagnostic label, but since it is simply the same process as already described by ALS, PMA and PLS, use of the term bulbar palsy implies that ALS, PMA and PLS refer to spinal forms of the disease, which is not the case. Furthermore, the term “bulbar palsy” merely refers to weakness involving bulbar-innervated muscles, which can be seen in those with any form of ALS, PMA or PLS, and varies in degree between people and with time. It is therefore best regarded as a phenotypic description rather than a diagnostic label. Since the trial of Riluzole in ALS in 1994, the use of the term “bulbar onset” has become more widespread and is arguably a better descriptor since it is clearly a phenotypic description rather than a possible diagnostic label, and does not change with time.{Bensimon, 1994 #80} Thus separating the diagnostic classification from phenotypic description would seem a logical way to proceed.
Clinical phenotype based on onset or progression
There are other possible ways to distinguish phenotypes. The risk an individual has developed ALS increases with age, reaching about 1 in 300 for men by age 80.{Johnston, 2006 #82} In other neurodegenerative diseases, age of onset distinguishes the underlying cause. For example, in Alzheimer’s disease, late onset disease is more likely to be apparently sporadic, and is a
heterogeneous syndrome related to vascular disease, inflammatory disease, protein deposition and APOE genotype, whereas young onset disease is more frequently associated with Mendelian disease gene mutations and a family history [A: ref?]. While people who develop ALS at a young age are more likely to be male and less likely to have bulbar onset disease,{Turner, 2012 #83} age of ALS onset has a log-normal distribution without a clear way to distinguish young and old. As a continuous phenotype, age of onset has no objective categories, making it difficult for existing classification systems to use.
Rate of progression has also been used to describe phenotype [A: ref?]. The advantage is that the prognosis is immediately apparent and it is plausible that rate of progression corresponds to the underlying cause. From a clinical perspective, although rate of progression is fairly constant and can be estimated from diagnostic delay or rate of decline in the ALS Functional Rating Scale,{Knibb, 2016 #393} at least two time points are needed to estimate it and the cut-off between fast and slow progression is not objectively defined.{Gordon, 2010 #84;Rooney, 2014 #336} Again, as a continuous phenotype, existing classifications that use categories cannot easily handle rate of progression. Any new system should ideally handle continuous phenotypes or allow a descriptor where the value of age of onset or rate of progression is deemed extreme or important enough to report.
Based on detailed and meticulous evaluation of patients longitudinally, the clinical manifestations of disease spread have been characterised and suggest that a pathogenic process occurs across contiguous regions of the neuraxis.{Ravits, 2009 #343} Although this work has never been replicated, the clinical observation is consistent with neuropathological staging based on the observed sequential pattern of TDP43 protein deposition, which appears to be propagated along axonal pathways.{Brettschneider, 2014 #347} Neither of the two main existing methods of
classification allow patterns of spread to be included to aid classification or as a descriptor, but such patterns may well be important for prognosis, stratification or genotype-phenotype correlation. [A: as above, for all these points, what are the implications for existing criteria/classification systems and for the development of new systems?**]
Classification of continuous phenotypes
A recurring problem is that ALS classification attempts to categorize continuous variables across which the relationships are complex and possibly heterogenous [A: ref?]. Phenotypes range from pure lower motor neuron involvement to pure upper motor neuron involvement, from aggressive disease to slowly progressive, from young onset to old onset, from symmetrical disease to
asymmetrical disease, and from a pure motor syndrome to ones with varying degrees of extra-motor involvement, with varying degrees of overlap and evidence of continuity from a clinical and
neuropathological perspective. (Table 1, Figure 2).
In an attempt to overcome these issues, researchers have turned to statistical models such as latent class cluster analysis [A: ref?**]. {Ganesalingam, 2009 #52} The biologically plausible idea is that underlying the clinical manifestations of ALS are hidden “latent” subgroups representing the different diseases that are collectively called ALS [A: ref?** same ref]. Using statistical clustering techniques, patterns of variables are found that can predict class membership. Using this approach, ALS has been shown to comprise five clusters, and even though prognosis was not used as a
classifying variable, the clusters are characterized by remarkable outcome differences.
{Ganesalingam, 2009 #52} Briefly, Class 1 (short survival) included mainly male patients with a classic or flail arm phenotype, Class 2 (short survival) included mainly female patients with bulbar
phenotype, Class 4 (long survival) spinal patients with an intermediate diagnostic delay, and Class 5 (long survival) mainly male patients with classic ALS or PLS phenotype but with a longer diagnostic delay and a young age at onset. Class 3 (very long survival) with an extremely long survival was limited to 4 patients in the original study. Such methods are useful research tools, but do not easily lend themselves to clinical practice in its current iteration, although this could change as additional technologies of clinical measurement based on genomics, MR and PET imaging and genomics become clinically accessible Hierarchical cluster analysis, based on clinical phenotypes can
discriminate 3 groups.{Elamin, 2015 #338} [A: which three groups? Please could you expand on this point to clarify the relevance of this method here?] Prognostic factors include site of onset of disease, slope of ALSFRS and presence of executive impairment. [A: is this for hierarchical cluster analysis?]
Forming a logical and useful classification of ALS based on continuously varying phenotypes is not straightforward, and the overlap between diagnosis, phenotype and progression further complicates the issue (Table 2). This difficulty is particularly obvious in the confusion of whether the El Escorial criteria are diagnostic for ALS or categorize the phenotype “ALS” within a broader spectrum of motor neuron disease. Despite revisions, this issue remains unresolved. [A: please ensure that these difficulties with the El Escorial criteria have been made clear in the above text** done and para reworded slightly to clarify]
The growing understanding of extramotor features of ALS, overlap with other neurodegenerative diseases and the identification of genetic and pathological subtypes of ALS add to the inadequacy of the El Escorial Criteria and mean that a further revision is inevitable.
Extramotor features in [A: diagnosis and?** yes] diagnosis and classification
One of the inadequacies [A: OK?** yes] of the El Escorial Criteria relates to the lack of a specific stipulation regarding the extra-motor features of ALS, i.e., frontotemporal dementia, oculomotor, autonomic nervous system, basal ganglia, cerebellar and sensory involvement in the diagnostic constellation of motor neuron disease/amyotrophic lateral sclerosis.{McCluskey, 2014 #307;Brooks, 2014 #308} Up to 13% of all incident ALS cases are associated with coincident frontotemporal dementia.{Phukan, 2012 #61} Indeed, extensive assessment of cognitive, behavioural and social cognitive function in ALS points to significant disruption of extra-motor systems [A: ref?** same ref Phukan]. Up to 50% of patients with ALS have clinical evidence of cognitive and behavioural decline, associated with changes in frontotemporal and frontostriatal pathways [A: ref?** same]. The presence of executive dysfunction is associated with more rapid disease progression and reduced survival [A: ref?** same]. Perhaps surprisingly, despite the wealth of deep phenotyping data, current diagnostic and stratification parameters do not include cognitive or behavioural status.
Notwithstanding, detailed examination of behavioural changes suggests that up to 40% of patients with ALS can be clustered into at least 5 different groups [A: what are these five groups?], which can be mapped onto known neuroanatomical pathways.{Elamin, 2013 #300}
The cognitive phenotype is a continuous phenotypic trait associated with the core motor features of ALS, ranging from normal cognition to overt dementia. However this continuous trait does not apply at an individual level, and we still do not know whether we should consider patients with ALS with frontotemporal dementia as a diagnostic category distinct from patients with ALS and normal cognition, or simply a phenotypic variant of ALS reflecting more widespread pathological burden of disease. Clues from neuropathology support different pathogenic mechanisms - for example, microglial activation and TDP43 pathology in frontotemporal dementia can be shown to correlate with neuropsychological measures of executive dysfunction, and there is evidence of different neuropathological patterns of TDP43 deposition in patients who present with behavioural variant frontotemporal dementia.{Brettschneider, 2012 #346;Brettschneider, 2014 #347} Similarly, new data from advanced imaging studies suggest differing patterns of disruption across structural and
functional connectomes at group level, implying that that the clinical manifestations of ALS are driven by selective network vulnerability. Imaging studies with MRI and 18FDG-PET seem to indicate that ALS patients with predominantly bulbar or spinal phenotypes have functional and structural differences.{Cardenas-Blanco, 2014 #49;Cistaro, 2012 #50;Prell, 2013 #62} Moreover, patients with different degrees of cognitive impairment show significantly different patterns of frontal lobe metabolic impairment on 18FDG-PET imaging.{Canosa, 2016 #305} Unfortunately, reliable biomarkers of upper motor neuron dysfunction and of cognitive impairment in ALS are not yet available for individual patient evaluation. Without them we are restricted to phenotypic
assessment, which is subjective and dependent on the extent of lower motor neuron involvement in the same limb for upper motor neuron dysfunction, or time consuming for cognitive assessment. Although a new diagnostic or classification system must include upper motor neuron involvement and cognitive impairment, accuracy will be affected by the ability to measure these features clinically. [A: please could you add a brief discussion of the implications for the development of a new classification or diagnostic system?** done]
Genetic findings in diagnosis and classification [A: I’ve edited this heading for length but your original question here is a very useful one and I’m sure readers would welcome some answers or discussion – please could you aim to critically evaluate the use of genetics in diagnosis and classification in this section?]
Family history
A standard classification of ALS has been to divide the disease into familial and sporadic disease, with up to 10% of cases described as familial. Although this might seem a simple dichotomy it is fraught with problems. First, the definition of what constitutes familial disease is imprecise and differs between neurologists.{Byrne, 2012 #276;Byrne, 2011 #277;Byrne, 2011 #278} For example, an affected cousin may not be regarded as constituting a positive family history, while an affected sibling would. Second, a family history of diseases other than ALS may be very important, but is not always taken, and until recently would not even have been recognised as relevant. For example,
frontotemporal dementia is not always accurately diagnosed, may be simply labelled “dementia” or may not be recognised as significant by an affected person. Other diseases (e.g. schizophrenia /psychosis in kindreds with the C9orf72 variant) may in the future be identified as also being more frequent in families with ALS but not be currently recognised as implying a positive family history. {Byrne, 2013 #131} Third, ALS is a disease of older age groups, and the causes of death of family members in previous generations may not be easily available.
These issues all bias the registration of a family history, but in addition, the concept itself is flawed. It cannot be argued that lack of a family history implies non-genetic disease because it is trivial to show that the probability of more than one affected depends on family size and disease gene penetrance. {Al-Chalabi, 2011 #172;Al-Chalabi, 2010 #177} When this is combined with the ascertainment biases and inconsistencies in classification it becomes clear that one would expect apparently sporadic cases to frequently carry Mendelian ALS-causing genes. In support of this, every reported familial ALS gene has been seen in people with no recorded family history,(http://alsod.kcl.ac.uk){Wroe, 2008 #210;Kenna, 2013 #306} and because reports of disease related variants are not always fully validated, the association of many variants with disease risk has not been fully established. Furthermore, the risk to relatives of those with sporadic ALS is increased,{Hanby, 2011 #170} and genes with variants of intermediate penetrance such as C9orf72 are detectable in family-based genetic studies{DeJesus-Hernandez, 2011 #279;Renton, 2011 #280} and in genome-wide association studies of those with apparently sporadic disease.{Shatunov, 2010 #178;van Es, 2009 #193} Any use of genetics in classification must therefore be based on pathogenic genetic variants identified in each individual, not on family history of disease.
Genetic contributions [A: OK?** yes]
Genetics does potentially provide an opportunity for a meaningful classification of ALS, because it carries the potential to subcluster various phenotypic manifestations into discrete underlying causes, regardless of the phenotype. This is potentially useful, as initial attempts at genotype-phenotype correlation for SOD1 mutation have not been particularly successful with the exception of variants such as p.A5V, which is usually associated with aggressive ALS, while those with homozygous p.D91A mutation tend to have very slowly progressive disease with a generally ascending upper motor neuron phenotype.{Andersen, 2011 #168} Other mutations do not have such a consistent phenotype, but this may also be because pathogenicity is assumed rather than known for most mutations. Thus, while the standard genetic classification of ALS1, ALS2 etc is in limited use, it is not helpful as a descriptor, carries no clinical information, and does not relate genotype to phenotype. SOD1 mutation was one of the first identified genetic causes for familial disease of any kind, and historically the extent to which private or rare mutations might be found in normal individuals was unknown. With the advent of high throughput whole genome sequencing it is now apparent that many SOD1 mutations attributed to ALS do not have strong supportive evidence.{Abel, 2013 #132;Kenna, 2013 #306} One fairly consistent theme is that SOD1 mediated ALS tends to show a predominantly lower motor neuron phenotype with no definitive evidence of cognitive impairment. Similarly, mutations in the FUS gene are generally characterised by an age at onset relatively younger (20 to 40 years) than for any other gene variants and a very poor outcome (less than 1 year from
onset).{Millecamps, 2010 #281} These mutations are often associated with a predominant weakness in neck and shoulder girdle muscles, in particular when caused by the mutation p.R521C.{Blair, 2010 #282} Other genes related to ALS cause a wide range of phenotypes even within the same pedigree, although mutations in some of them are more frequently associated with a lower motor neuron phenotype (FUS, SOD1, CHCHD10, MATR3, VCP) and others with upper motor neuron signs (Alsin, UBQLN2, FIG4).{Andersen, 2011 #168} The C9orf72 expansion appears to have an overlap with psychiatric disorders.{Fahey, 2014 #283;Kertesz, 2013 #284}
Mutations in many ALS-related genes are characterized by ALS and frontotemporal features in a varying percentage of patients, the major exception being mutation of SOD1, which is characterized by spared cognitive function.{Wicks, 2009 #199} C9ORF72 expansion is probably the genetic
mutation most frequently associated with cognitive impairment.{Hardy, 2014 #53}
The observation that genetic background strongly influence the phenotypic presentation in transgenic mouse models of ALS points toward the hypothesis that clinical phenotypes may have a biological basis also in humans.{Heiman-Patterson, 2005 #54;Heiman-Patterson, 2011 #55} In particular, there appears to be a possible disconnect with respect to the clinical course of ALS in different ethnic groups as a function of their location{Rojas-Garcia, 2012 #163;Kazamel, 2013 #310;Gundogdu, 2014 #309} compared with their place of origin{Marin, 2015 #311;Nalini, 2008 #312} despite strong evidence for other disease features including risk, age of onset and incidence to be clearly affected by ethnicity.{Rechtman, 2015 #313;Mehta, 2014 #314;Zaldivar, 2009 #315} At present, our limited knowledge of genetic relationships to biological phenotypes in ALS suggests that genes should be included in a classification system, but their meaning may not be apparent until our understanding has improved.
Genotype-Environmental Interactions
One emerging area of study that will impact the classification of ALS is the recognition that environmental factors might not only be causative in the development of ALS but could also affect the progression of the disease.{Barbeito, 2010 #316;Kamel, 2008 #317} Genetic studies show that the effects of modifying genes on progression may depend on an environmental factor potentially involved in the causation of the ALS phenotype.{Keren, 2014 #114;Eum, 2015 #318;Fogh, 2016 #394} It may therefore become appropriate to include environmental descriptors in a classification system, but at present there are no widely accepted environmental causes of ALS.
Pathological subtypes in classification [A: OK?** yes]
In cancer, a cornerstone of classification is pathology. In ALS, as in many other neurological disorders, [A: addition OK?** yes] the situation is complicated by the lack of a definitive tissue area for biopsy. The most important finding is an inclusion type labelled by antibodies to TDP43, seen in 98% of people with ALS at post mortem[A: at post mortem?** yes - added] (such inclusions are not reported in those with SOD1 mutation, or in those with FUS mutation).{Neumann, 2006
#285;Kwiatkowski, 2009 #286;Al-Chalabi, 2012 #150;Mackenzie, 2007 #287;Vance, 2009 #200} This finding suggests that the final common pathway for most ALS involves TDP43, but that at least two other pathways lead to the same phenotype through a different mechanism. Conversely, TDP43 pathology is also well-recognised in other neurodegenerative conditions, including Alzheimer’s disease [A: ref?]. The observation of TDP43 pathology in Alzheimer’s disease, but also in cognitively normal older individuals [A: ref?] may suggest the existence of a pathological threshold that differentiates a preclinical phase prior to the manifestation of an amyotrophic lateral sclerosis (or dementia) clinical syndrome. Whether this is defined by the regional distribution or burden of TDP43 pathology has not yet been established.
The emergence of multiple staging schemes for the progression of TDP43 pathology in amyotrophic lateral sclerosis, and also behavioural variant frontotemporal dementia and Alzheimer’s disease, may suggest overlap in the vulnerable regions implicated in the different clinical syndromes[A: ref?] . {Brettschneider, 2013 #319} However, to date, there have been few ALS studies to determine
whether regional concentration of TDP43 pathology is relevant to specific clinical phenotypes. While pathological burden measured by TDP43 pathology has been proposed to correlate with clinical changes in the brain and spinal cord, and the presence of microglial activation may correlate with the presence of executive impairment,{Brettschneider, 2012 #346} recent observations have identified that the pathological burden in advanced ALS may include the oculomotor areas of the brainstem, supporting the concept that the pathological burden may increase in regions of the central nervous system previously considered to represent ALS-Plus.{Brettschneider, 2014 #320} Indeed, clinical studies have identified various levels of oculomotor dysfunction that might be crucial to the
classification of ALS in the future.{Gorges, 2015 #321} Complicating these future investigations is the fact that TDP-43 pathology may be influenced by ethnic background.{Nascimento, 2015 #322} Because pathological findings are only accessible after death in ALS, they cannot be used in a clinical classification, but could be added to a pathological classification, and can inform approaches to clinical classification, such as whether spread of disease should be included, and whether ALS with TDP43 inclusions should be explicitly separated from other forms of ALS.
[A: as for the previous sections, please could you expand on the implications of the pathological subtypes for classification systems?** done]
Biomarkers and laboratory subtypes in diagnosis and classification [A: OK?** yes]
Diagnostic markers for ALS are an important subject of research. A number of laboratory tests have been assessed for their suitability to identify ALS mimics{Alanazy, 2014 #323;Diagnosis, 2012 #324;Brooks, 2000 #328;Sanderson, 2015 #329} [A: edited to clarify that these are still in the research stage – OK?** yes] but it is not clear if the presence of antibodies or other laboratory abnormalities in the context of ALS provide an avenue for treatment that might affect the course of disease.{Donaldson, 2016 #330;Kollewe, 2015 #331} Just as genotype effects may be both positive and negative with regard to disease progression, there is the potential for immune reactions in ALS to potentially modify disease progression.{Ehrhart, 2015 #332;Murdock, 2015 #333;Tzartos, 2014 #334;Hwang, 2013 #335}
Serum levels of light chain neurofilaments (NfL) were recently shown to have >90% sensitivity and specificity in distinguishing ALS from healthy controls,{Gaiottino, 2013 #348} but more importantly, CSF levels of NfL had a 77% sensitivity and 85% specificity in separating ALS from ALS-mimic disorders.{Steinacker, 2016 #349} CSF phosphorylated heavy chain neurofilaments showed similar sensitivity and specificity. The immunoreactivity to serum NfL was significantly related to ALS clinical staging,{Puentes, 2014 #350;Roche, 2012 #161} indicating that this biomarker may also be sensitive to disease progression.
MRI can help distinguish between ALS and non-ALS forms of MND. PLS compared to ALS shows more severe damage to the motor corpus callosum fibres and subcortical white matter (WM) underlying the primary motor cortices; furthermore, in PLS the damage to the callosal mid-body correlates with the severity of upper motor neuron clinical burden.{Agosta, 2015 #46} Conversely, no differences in cortical thickness is seen between patients with lower motor neuron-predominant disease and those with ALS confirming the concept of the frequent subclinical involvement of upper motor neuron pathology in subjects with a predominant lower motor neuron phenotype.{Spinelli, 2016 #351} Diffusion tensor imaging has been used to identify ALS-associated sequential regional patterns of brain alterations at the group level, and correlates with pathological spread {Brettschneider, 2014 #347;Brettschneider, 2013 #319} and ALS functional rating scale scores.{Kassubek, 2014 #352} Biomarkers have the potential to radically alter our classification of ALS. If reliable diagnostic markers can be found, or sets of markers that correlate with clinical phenotypes or genetic subtypes
identified, they would be a quick route to classification, and potentially more personalised therapy, targeted at the specific biological pathway represented. [A: as before, please could you add discussion of the implications of biomarkers and laboratory subtypes for classification and diagnostic systems?** done]
Where do we go from here?
Discussions as to how best to update the El Escorial diagnostic criteria to account for new insights (eg advances in neurophysiology, the presence of cognitive or behavioural impairment in ALS, the overlap with frontotemporal dementia, an increased understanding of genomics) are currently underway within the World Federation of Neurology, with the clear recognition that the criteria are primarily for research purposes and clinical trial enrolment rather than for use in clinical practice. The first suggestions have already been published.{Ludolph, 2015 #344;Agosta, 2015 #46} [A: when are these updates expected and what, briefly, is the process?** slightly modified to clarify. Please also see my e-mail about a possible panel on this ** could be separated into a panel from the start of this paragraph to end of next paragraph.]
Moreover, there is now an opportunity to expand the ontology of disease phenotype in ICD-11 as it applies to individual patients (Table 3) providing an enriched broader spectrum of phenotype information that might lead to better integration of biomarker and genetic markers of disease progression.{Treede, 2015 #294;Rajakulendran, 2014 #295;Rief, 2010 #296;Jette, 2015 #297;Stone, 2014 #298} The elements detailed in the Airlie House Revision of the El Escorial criteria, and its appendices{Brooks, 2000 #4} need to be attended to in order to expand the known phenotype ontology that has developed in the last decade in terms of signs, symptoms, laboratory findings and imaging studies to allow linkage with disease biomarker and disease genetic markers or modifiers. {Ganesalingam, 2009 #52;Statland, 2015 #299;Statland, 2015 #35;Liewluck, 2015 #36;Jawdat, 2015 #37} The Human Phenotype Ontology algorithms (http://human-phenotype-ontology.github.io/) [A: please explain what these are or provide an online link** done] might allow the World Federation of Neurology Research Group on Motor Neuron Diseases to provide a more appropriate framework to the currently proposed classification of motor neuron diseases in ICD-11.{Groza, 2015
#38;Brookes, 2015 #39}
The time is ripe for the motor neuron disease community to engage in the development of the final structure of the ICD-11 9D00 Motor Neuron Diseases and Related Disorders. The process entails refining the clinical descriptions and diagnostic guidelines in ICD-11{First, 2015 #44} and then subjecting these to field tests to define their content validity, reliability, and usefulness. For example, as part of the United States National ALS Register conducted by the Agency for Toxic Substances and Disease Registry (ATSDR), the Airlie House revision of the El Escorial criteria were field tested and found to be robust in diagnostic processes for epidemiological studies.{Wagner, 2015 #45} Sifting and winnowing of the innovative classification proposed will have to be an ongoing process focused on a working model by 2018.{Stein, 2016 #353}
On the research side, [A: addition OK?** yes] a radical and completely novel approach towards understanding disease pathogenesis could be achieved by exploiting the presence of disease heterogeneity at a multidimensional level. Genotype phenotype studies, recognizing the likely importance of previously ignored genetic pleiotropy, could enable development of recent discoveries of overlapping pathways in neurodegeneration and neuropsychiatric disease.{Byrne, 2013 #131} This re-orientation could result in innovative approaches towards understanding disease using extensive population based phenotyping, detailed genealogical studies and modern genomics.
The approach could be enhanced using the concept of network failure, and development of novel disease signatures based on integration of clinical, genomic, transcriptomic, neuroimmune, imaging
and network analysis data [A: refs?]. Such a model could provide previously unrecognized sub-cohorts, suitable for inclusion in “precision medicine” based trials. Indeed, this could in turn open the field to development of drugs with multimodal mechanisms of action, moving the landscape of ALS and related neurodegenerations towards the modern era of personalized medicine. Other approaches to causation, such as the multistep model of ALS,{Al-Chalabi, 2014 #109} may lead to classification systems in the future as our understanding of disease pathways improves. (See Panel) [A: please see notes in the e-mail about this panel**]As a first step, we propose as standard, a more nuanced diagnostic description of ALS that incorporates in order: clinical stage, the balance between upper and lower motor neuron involvement (default is typical ALS) within a diagnostic category, relevant risk factor or biomarker (optional), El Escorial certainty of ALS phenotype (if ALS), and any striking phenotypic features (figure). The diagnostic category can be modified by the addition of the terms familial or FTD. For example: “Stage 2, upper motor neuron predominant ALS, FUS mutation, fulfilling the El Escorial Probable criteria, bulbar onset”, or “Stage 3 familial ALS-FTD, C9orf72 expansion, El Escorial definite, young onset”. We further propose that the terms PMA and PLS are removed from the diagnostic component unless more than two years have passed since symptom onset.
[A: please could you clarify for the readers how these different diagnostic descriptors relate to each other – eg, are you proposing this as a hierarchy of diagnostic categories? If so, perhaps arrows between the categories or numbers for the different components would help to give clinicians an order or staged process to follow. Arrows from the diagnostic modifiers and optional terms to the categories they affect or some other addition to ensure that these different
components hang together might also be helpful** Legend below. Figure separate]
This system has the advantage of providing key information in a systematic way, while allowing individual preferences for phenotypic descriptors. It is flexible enough to incorporate future
discoveries, and allows a personalized approach because it provides significant detail of both cause, balance of upper and lower motor neuron involvement, and clinical stage. The last field on
phenotypic descriptors can also be modified if new classifications come to light, and allows cognitive involvement to be explicitly stated.
Adoption by ALS practitioners should be straightforward, since most of the components are widely used already. A recommendation to use this more complete classification from a recognised body, such as the World Federation of Neurology Research Group on Motor Neuron Diseases, or the American Academy of Neurology, would standardize practice. Alternatively, adoption by members of the European Network for the Cure of ALS, ENCALS (http://www.encals.eu/) or the North East ALS consortium, NEALS (http://www.alsconsortium.org/), or recommendation from a patient body such as the MND Association (http://www.mndassociation.org/) or ALS Association (http://www.alsa.org/) would raise awareness and speed implementation by ALS practitioners. Regular review by the World Federation of Neurology Research Group on Motor Neuron Diseases would ensure that changes to the system could be made if needed, to ensure it remains up to date and relevant.
[A: what is needed to take this system forward? What sort of testing or confirmation? What challenges are there for the implementation of such a system?**]
Conclusion
There are two major purposes to disease classification: clinical management and research. For research purposes, a detailed subclassification of ALS is important because new drugs targeted at the specific subtype can be developed using a precision medicine approach. Experience to date has been
that for potential neuroprotective and disease-modifying therapies to work, these therapies are best introduced at the earliest stages of the disease, which in turn must be driven by disease classification systems. For clinical management, [A: deleted because I’m not sure this comparison is helpful here] a detailed classification of the forms of ALS is of great importance to drug discovery and future treatment, but for every day diagnosis and management, a simpler overview may be more useful. [A: please could you end with a few sentences or a short paragraph to briefly summarise the next steps needed in the development of new diagnostic and classification systems so that you end with a forward-looking perspective?** done]
A dramatic change in our thinking is needed. It is no longer sufficient to muddle along with our own internal and inconsistent classifications, or to rely on incomplete or hierarchical systems that are restricted in purpose, such as ICD-10 or El Escorial criteria. We are at the advent of a new molecular understanding of ALS, and a modern, useful, clinically relevant, consistent classification is urgently needed for our patients and to take research forwards. We propose this new classification system as a start, allied with more research into the biology of ALS and the corresponding clinical
manifestations of ALS and related disorders. A robust classification system reflects a robust understanding of the underlying disease. We are now at a stage where, for the first time, this is possible for ALS.
Conflict of interest statement
AAC is Chief Investigator of a clinical trial by OrionPharma , [A: deleted as these are not considered as conflicts] has consulted for GSK, OrionPhama, Biogen Inc, Cytokinetics, and Treeway, and receives royalties from the books “The Brain” (Oneworld Publications) and “The Genetics of Complex Human Diseases” (Cold Spring Harbor Laboratory Press). OH has received speaking honoraria from Janssen Cilag, Biogen Idec, Sanofi Aventis, Novartis and Merck-Serono, and has been a member of advisory panels for Biogen Idec, Allergen, Ono Pharmaceuticals, Novartis, Cytokinetics and Sanofi Aventisa. AC serves on scientific advisory boards for Biogen Idec, Cytokinetics and Italfarmaco. BRB is Principal Investigator of a clinical trial by Medicinova [NCT02238626] and co-investigator in clinical trials by Cytokinetics [NCT02496767] and Alexion [NCT01997229]. LvdB received travel grants and
consultancy fees from Baxalta, and serves on scientific advisory boards for Prinses Beatrix Spierfonds, Thierry Latran Foundation, Cytokinetics and Biogen. MK declares no competing interests.
Author contributions
AAC and OH drafted the manuscript outline. All authors contributed to the draft manuscript and agreed the final submitted version.
Acknowledgments
AAC, LvdB, AC and OH are funded through EU Joint Programme - Neurodegenerative Disease Research (JPND) projects. The projects (SOPHIA, STRENGTH, ALS-CarE) are supported through the following funding organisations under the aegis of JPND - www.jpnd.eu (United Kingdom, Medical Research Council and Economic and Social Research Council, Netherlands, ZonMW, Italy, Ministero dell'Istruzione, dell'Università e della Ricerca, Ireland, Irish Health Board). AAC receives salary support from the National Institute for Health Research (NIHR) Dementia Biomedical Research Unit at South London and Maudsley NHS Foundation Trust and King’s College London. Work leading up to this publication was funded by the European Community’s Health Seventh Framework Programme (FP7/2007–2013; grant agreement number 259867). AAC has grants from the Motor Neurone Disease Association, ALS Association, National Institute for Health Research, European Commission, Medical Research Council and Economic and Social Research Council. AC has grants from Italian Ministry of Health (Ricerca Finalizzata), University of Turin, and Fondazione Vialli e Mauro onlus. MK is supported by funding to Forefront, a collaborative research group dedicated to the study of frontotemporal dementia and amyotrophic lateral sclerosis, from the National Health and Medical Research Council of Australia Program Grant #1037746. OH receives funding from the ALS
Association, HRB (the Health Research Board, grant H01300), Science Foundation Ireland Research Motor Neuron. BRB is supported by has research funding from Muscular Dystrophy Association, ALS Therapy Alliance and ALS Association, National Institute of Neurological Diseases and Stroke through NorthEast ALS Consortium, Massachusetts General Hospital, Harvard Medical School and Johns Hopkins University School of Medicine, Muscular Dystrophy Association ALS Division, Carolinas ALS Research Fund, Pinstripes Fund, Carolinas Garden of Hope, Heineman Medical Research Fund, and Carolinas Healthcare Foundation. LvdB received grants from the Netherlands Organization for Health Research and Development (Vici scheme), the Netherlands ALS Foundation, and the Prinses Beatrix Fonds.
Figure and table legends
Figure 2. The natural history of ALS
[A: please could you divide this figure and the legends into figure parts (A, B, C etc) and make sure that the legend clearly explains each figure part so that, as above, this figure can stand alone without reference to the main text? Please include in the legend only information that is essential for the readers’ understanding of the figure and remove any detail that is already stated in the main text] Genetic and environmental factors contribute to the onset of ALS, with disease onset occurring an unknown time before symptom onset. Clinical diagnosis typically takes about a year in most health systems. [A: should this figure part be drawn as a graph with time on the x-axis and disease progression on the y-axis? Perhaps the other parts (“Causes”, “Underlying pathological diagnosis”, and the “disease onset” and other labels) could then be superimposed onto this graph, so that the relationships between these stages etc are clear] Disease progression is relentless through different stages with increasing symptom burden and worsening function as measured by the ALS Functional Rating Scale (ALSFRS-R). The distribution of symptoms varies but eventually involves most or all voluntary muscles and may involve various extramotor systems, including frontotemporal degeneration. The age of onset is variable but ALS is rare below the age of 30. Survival is variable, but 50% of people are dead within 2 years of first symptoms. The El Escorial classification is closely related to disease stage and is used to provide a diagnostic and phenotypic test for ALS. [A: what do the blue and red lines in this figure part represent? And the pink shading?] Upper and lower motor neuron signs are required for a diagnosis of ALS but these vary with time, and may mask each other. For example, severe wasting and weakness resulting from LMN
involvement may make it difficult to detect the increased tone and brisk reflexes corresponding to UMN signs. [Please provide labels for both the x and y axes for all figure parts showing graphs (the current x-axis labels seem to be headings for the graphs)].
Figure X. The proposed classification of ALS
In clinical practice, or in clinical trials, the full description of ALS should be given, beginning on the left of the figure and proceeding through each block. The final block labelled “Phenotype” is flexible and corresponds to commonly used internal classifications that neurologists or other practitioners apply to characterize remarkable features of a particular phenotype. The set of terms labelled “Mandatory terms” are the essential component of description, with modifiers and optional terms added where needed. The exact order of blocks can be altered if needed, to make communication easier. For example, “Stage 2 Young Onset ALS fulfilling El Escorial Probable criteria” is equivalent to “Stage 2 ALS, El Escorial Probable, Young Onset.”
Table 1. Continuous variables forced into categories in the phenotypic classification of ALS Table 2. Descriptors used in the diagnosis and phenotypic classification of ALS
