Escolar Documentos
Profissional Documentos
Cultura Documentos
0000000000006876
ARTICLE CLASS OF EVIDENCE
®
Neurology 2019;92:e536-e547. doi:10.1212/WNL.0000000000006876
Objective Editorial
The aim of the study was to assess the utility of a novel amyotrophic lateral sclerosis (ALS) Upper motor neuron
diagnostic index (ALSDI). assessment and early
diagnosis in ALS: Getting it
Methods right the first time
A prospective multicenter study was undertaken on patients presenting with suspected ALS. Page 255
The reference standard (Awaji criteria) was applied to all patients at recruitment. Patients were
randomly assigned to a training (75%) and a test (25%) cohort. The ALSDI was developed in MORE ONLINE
the training cohort and its diagnostic utility was subsequently assessed in the test cohort.
Class of Evidence
Criteria for rating
Results therapeutic and diagnostic
A total of 407 patients were recruited, with 305 patients subsequently diagnosed with ALS and studies
102 with a non-ALS mimicking disorder. The ALSDI reliably differentiated ALS from neu-
NPub.org/coe
romuscular disorders in the training cohort (area under the curve 0.92, 95% confidence interval
0.89–0.95), with ALSDI ≥4 exhibiting 81.6% sensitivity, 89.6% specificity, and 83.5% diagnostic
accuracy. The ALSDI diagnostic utility was confirmed in the test cohort (area under the curve
0.90, 95% confidence interval 0.84–0.97), with ALSDI ≥4 exhibiting 83.3% sensitivity, 84%
specificity, and 83.5% diagnostic accuracy. In addition, the diagnostic utility of the ALSDI was
confirmed in patients who were Awaji negative at recruitment and in those exhibiting a pre-
dominantly lower motor neuron phenotype.
Conclusion
The ALSDI reliably differentiates ALS from mimicking disorders at an early stage in the disease
process.
Classification of evidence
This study provides Class I evidence that for patients with suspected ALS, the ALSDI distin-
guished ALS from neuromuscular mimicking disorders.
From Westmead Clinical School (N.G., P.M., M.v.d.B., S.V.), Brain and Mind Center (J.H., K.S., J.M.M., S.B.P., M.C.K.), and NHMRC Clinical Trials Centre (K.B.), University of Sydney; and
Westmead Hospital (K.B.), Research and Education Network, Sydney, Australia.
Go to Neurology.org/N for full disclosures. Funding information and disclosures deemed relevant by the authors, if any, are provided at the end of the article.
Amyotrophic lateral sclerosis (ALS) is a progressive neuro- subclinical UMN dysfunction as a diagnostic biomarker of
degenerative disorder of the human motor system, charac- ALS. In addition, it has been postulated that cortical hy-
terized by a progressive loss of upper motor neurons (UMNs) perexcitability is an adverse prognostic biomarker in ALS,29
and lower motor neurons (LMNs) (and interneurons), evolving with disease progression.30 Although TMS di-
resulting in initial dysfunction and later total loss of function agnostic criteria for diagnosis of ALS have been previously
of the aforementioned neurons.1,2 The diagnosis of ALS reported,26 these have not been validated in independent
remains reliant on identification of concurrent UMN and cohorts.31 In addition, the current ALS diagnostic process
LMN signs,3–5 along with evidence of disease progression and is complex, relying on the identification of numerous
exclusion of neuromuscular mimicking disorders.2,6 In the clinical and neurophysiologic abnormalities. Consequently,
absence of a pathognomonic test and a heterogeneous clinical the aim of the present study was to use multivariable sta-
phenotype, significant diagnostic delays have been reported.7–9 tistical analysis to develop a diagnostic score for ALS in-
Consequently, institution of appropriate management strategies, corporating clinical, conventional neurophysiologic and
including neuroprotective therapies and recruitment into clinical TMS measures available on recruitment, termed the ALS
trials may be critically delayed, perhaps beyond the putative diagnostic index (ALSDI), in order to reliably differentiate
therapeutic window period.10,11 ALS from neuromuscular disorders. The ALSDI test was
developed in the randomly selected training cohort and its
A greater understanding of ALS pathophysiology has led to performance subsequently assessed in the independent
identification of novel therapeutic approaches,2 and the validation test cohort to minimize potential bias.32
consequent need for efficient clinical trial designs. Despite
this need, the diagnostic algorithm remains complex, resulting
in uncertainty in requirements for patient recruitment.
Methods
Attempts to develop more efficient ALS diagnostic criteria led Study design and patients
to the first international workshop resulting in the clinically In this prospective study undertaken according to the STARD
based El Escorial criteria.3 Subsequently, the neurophysio- (Standards for Reporting of Diagnostic Accuracy Studies)
logically based Awaji criteria were developed as an add-on to criteria, potential participants were identified following de-
the El Escorial diagnostic criteria.5 The Awaji criteria exhibi- tailed clinical and neurophysiologic assessment at 2 neuro-
ted greater sensitivity,12–20 but the diagnostic benefits were muscular centers integrated through Sydney Health Partners,
most prominent in bulbar-onset ALS.16 Incorporation of the University of Sydney. Subsequently, patients were consecu-
“clinically probable laboratory-supported” diagnostic cate- tively and prospectively recruited in accordance with the in-
gory, necessitating the identification of UMN and LMN clusion criteria. The investigators were blinded at the time of
dysfunction in one region, increased the sensitivity of the enrollment to the final diagnosis, which was established after
Awaji criteria.21 More recently, a further revision of the El at least 6 months of follow-up. The investigators that per-
Escorial criteria was proposed, whereby identification of formed TMS testing were blinded to the eventual diagnosis
UMN and/or LMN signs in one region was considered di- at TMS assessment. The inclusion criteria were as follows:
agnostic.22 Given the reliance of clinical identification of (1) pure LMN syndrome, (2) mixed UMN and LMN syn-
UMN dysfunction and difficulties in identifying UMN signs drome, or (3) UMN syndrome. Patients were recruited early
in patients with ALS,23 the sensitivity of these diagnostic in the disease process, with the median duration of disease
criteria could be enhanced by objective assessment of the onset in the ALS cohort 12 months (interquartile range 7–20
corticomotoneuronal system. months).
Transcranial magnetic stimulation (TMS) techniques enable The exclusion criteria included the following: (1) use of psy-
an objective assessment of the UMN functional integrity in chotropic medications that could potentially affect TMS
human CNS. Of importance, cortical hyperexcitability was measures; (2) acute migraine headache within 4 weeks pre-
identified as an early and specific feature in ALS, clearly dif- ceding assessment; (3) history of other neurologic diseases
ferentiating ALS from neuromuscular mimic disorders.24–26 including movement disorders, epilepsy, stroke, or TIA; (4) in
Separately, cortical hyperexcitability was reported in atyp- situ pacemakers and other cardiac devices, cochlear implants;
ical ALS phenotypes,27,28 underscoring the importance of (5) history of brain or cerebrovascular surgery; (6) patients
The tree includes the UMN score along with other continuous covariates and categorical variables, including the site of onset and inexcitability status. The
UMN score has replaced some of the other covariates, but it should be noted that this variable was systematically missing more often in the non-ALS group.
The inexcitability status and averaged SICI were of clear value in both the training and test sets. The overall performance in the training cohort was 89.6%.
Adj. = adjusted; ALS = amyotrophic lateral sclerosis; df = degrees of freedom; SICI = short-interval intracortical inhibition; UMN = upper motor neuron.
The tree includes the UMN score along with the other continuous covariates and the categorical variables site of onset and inexcitability status. The UMN
score has replaced some of the other covariates, but it should be noted that this variable was systematically missing more often in the non-ALS group. The
inexcitability status and averaged SICI were of clear value in both the training and test sets. The overall performance in the test cohort was 86.8%. Adj. =
adjusted; ALS = amyotrophic lateral sclerosis; df = degrees of freedom; SICI = short-interval intracortical inhibition; UMN = upper motor neuron.
In the same sitting, the degree of LMN dysfunction was de- Before developing a diagnostic index for ALS at final diagnosis
termined by electrically stimulating the median nerve at the wrist based on clinical and neurophysiologic measures assessed at
and recording the peak-peak CMAP amplitude, distal motor recruitment, SPSS was used to split the dataset at random
latency, F-wave latency, and frequency. The neurophysiologic in a 3:1 ratio into the training cohort (approximately 75%
index was determined using a previously reported formula.36 of study participants) and the remaining independent test
cohort (approximately 25%). The 3:1 ratio was deemed
Reference standard optimal to assess the accuracy of the ALSDI based on ad-
The reference standard was the Awaji diagnostic criteria.5 A vice from the biostatistician (K.B.). The ALSDI was de-
positive reference standard was regarded as classification into veloped in the training cohort. The receiver operating
a definite, probable, or possible category based on assessment characteristic curve for ALSDI in the training cohort was
at recruitment (first visit). used to determine an appropriate cutpoint for the index
test. The diagnostic performance of ALSDI was sub-
Statistical methods sequently assessed in the independent test cohort to min-
The software IBM SPSS Statistics version 23 (IBM Corp., imize potential bias.32
Armonk, NY) was used to analyze the data. Continuous
variables were summarized using mean and SD or median and The primary outcome measure was the diagnostic utility of
lower quartile to upper quartile as appropriate. Frequencies or ALSDI in differentiating ALS from non-ALS neuromuscular
percentages were used to summarize categorical variables. disorders in the test cohort. The area under the receiver op-
Chi-squared tests (or exact permutation tests if appropriate) erating characteristic curve (AUC) was used to quantify the
were used to test for univariable associations between cate- diagnostic utility of ALSDI. Sensitivity, specificity, positive
gorical variables. Two-tailed tests with a significance level of and negative predictive values, diagnostic odds ratios (ORs),
5% were used throughout. and accuracy for particular ALSDI cutpoints, together with
After the study was closed and all patient data had been collated, the dataset was split at random in a 3:1 ratio into the training cohort (approximately 75% of
study participants) and a test cohort (approximately 25%). The index test was developed in the training cohort and its performance then assessed in the
independent test cohort. In this flow diagram, a positive index test was defined as ALSDI ≥4. ALS = amyotrophic lateral sclerosis; ALSDI = amyotrophic lateral
sclerosis diagnostic index.
Table 1 Clinical and demographic features of patients with ALS and neuromuscular mimic disorders
Neuromuscular ALS mimic disorders (n = 102) ALS excitable (n = 248) ALS inexcitable (n = 57)
Age, y 53 40 63 62 54 70 62 48 69
MRC sum 90 86 90 82 76 87 84 71 89
UMN score 0 0 0 10 3 12 13 12 14
Abbreviations: ALS = amyotrophic lateral sclerosis; MRC = Medical Research Council; UMN = upper motor neuron.
All data are expressed as mean (SD) or median (interquartile range). The neuromuscular mimic disorder group included acquired neuromyotonia syndrome
(26), hereditary spastic paraplegia (12), myopathy (12), Kennedy disease (9), Hirayama disease (8), pure motor chronic inflammatory demyelinating neu-
ropathy (7), multifocal motor neuropathy (7), spinal muscular atrophy (6), FOSMN (facial-onset sensory and motor neuronopathy) syndrome (4), distal
hereditary motor neuronopathy with pyramidal features (3), postpolio syndrome (2), cervical radiculopathy (2), Pompe disease (2), lumbosacral radiculopathy
(1), and lead toxicity (1). A majority of patients classified as “myopathy” were diagnosed with inclusion body myositis (75%). All patients were assessed using
the MRC and UMN scores. The UMN score was not available for 15% of patients without ALS and 2.6% of patients with ALS.
Age at assessment, y
<50R
Limb
R
>9.01
CSP duration, ms
>220.1R
Mean SICI, %
R
>12.1
Abbreviations: CI = confidence interval; CMAP = compound muscle action potential; CSP = cortical silent period; R = reference; SICI = short-interval intracortical
inhibition.
Five independent predictors were identified, namely, age at assessment, site of disease onset, median nerve CMAP amplitude, CSP duration, and mean SICI.
a
Significant.
compared to the Awaji criteria (figure 4C), although the dif- specificity, diagnostic OR 42.3, and 87.2% accuracy compared
ference was not significant as indicated by overlapping to 82.5%, 70.6%, 11.3, and 81.8%, respectively, in patients
95% CIs. who were Awaji positive (table 4). Of note, application of the
index test (ALSDI ≥4) in patients initially classified as Awaji
From the above it was determined that the sensitivity and negative (n = 133) resulted in identification of an extra 38
diagnostic accuracy were higher for both cohorts when an patients with ALS (29%) at the risk of misclassifying 7 non-
ALSDI cutpoint of 4 was used. Consequently, in the subgroup ALS patients as ALS.
analyses, based on Awaji positive versus negative status at
recruitment, and prominent UMN signs (UMN score >3) The diagnostic utility of the index test (ALSDI ≥4) was also
versus paucity of UMN signs (UMN score < 3), we therefore established in predominantly LMN phenotypes of ALS.
assessed the diagnostic utility of ALSDI ≥4. Specifically, ALSDI ≥4 had comparable sensitivity in patients
expressing a paucity of UMN signs (UMN score <3) and in
An ALSDI ≥4 exhibited a comparable sensitivity and di- those with prominent UMN signs (UMN ≥3) (table 4). It
agnostic accuracy in patients initially classified as Awaji neg- should be stressed that the UMN score was not recorded in
ative and positive (table 4). Specifically, in patients who were 15% of patients without ALS and 2.6% of patients with ALS,
Awaji negative, ALSDI ≥4 exhibited 79.2% sensitivity, 91.8% and these patients were excluded from the analysis. Of
Age at assessment, y In total, 279 patients with ALS (68.6%) were receiving rilu-
zole at the time of assessment. Given that riluzole modulates
<50 0
SICI,39,40 a major component of the ALSDI, we assessed
51–60 1 whether riluzole therapy could reduce the diagnostic utility of
61–70 1
the index test ALSDI ≥4. The sensitivity was similar between
patients on riluzole (83.5%, 95% CI 76.0%–89.3%) and off
>71 1 riluzole (81.5%, 95% CI 74.3%–87.4%) therapy, thereby in-
Site of disease onset dicating that treatment status did not influence the diagnostic
potential of the ALSDI ≥4.
Bulbar 2
Limb 0
estimate of index test performance in a real-world setting.45 which the SICI is normal, as evident in approximately 30% of
The robust diagnostic performance of ALSDI ≥4 in the patients with ALS.26
independent test cohort corroborated its diagnostic utility.
It should be stressed that an ALSDI ≥4 exhibited a speci- Of further relevance, the presenting clinical phenotype, pa-
ficity of 84%, while an ALSDI ≥5 had a specificity of 96%. tient demographics, and site of disease onset are important
Consequently, ALSDI scores <4 would be less appropriate considerations in the diagnosis of ALS. Incorporating clinical
for clinical trials. and peripheral neurophysiologic measures into the diagnostic
model, along with objective biomarkers of UMN dysfunction,
The present study utilized multivariable statistical analysis to could potentially enhance the diagnosis of ALS. In the present
identify independent biomarkers of UMN dysfunction. In study, an ALSDI ≥4, derived by summing age at onset (older
addition to abnormalities of SICI and motor cortex inex- than 50 years), low CMAP amplitude (<5 mV), and bulbar-
citability, reduction of CSP duration was also identified as an onset disease, reliably differentiated ALS from neuromuscular
independent diagnostic biomarker of ALS. While CSP re- mimicking disorders. An exclusive reliance on TMS abnor-
duction has been well established in ALS,26–28,46,47 a finding malities may preclude the diagnosis of ALS in patients with
attributed to dysfunction of long latency inhibitory circuits severe wasting of the target muscle (approximately 6% of ALS
acting via GABAB (γ-aminobutyric acid type B) receptors, cohorts).26 Consequently, the ALSDI could potentially en-
the diagnostic utility of this biomarker has not been pre- able a diagnosis of ALS in cases in which the target muscle is
viously utilized. In the present study, we demonstrated markedly wasted precluding TMS assessment. ALS diagnosis
a modest diagnostic potential of CSP duration (<160 milli- may also be delayed in predominantly LMN phenotypes. In
seconds), potentially aiding ALS diagnosis in situations in the present study, the utility of the index test ALSDI ≥4 was
Training cohort
Awaji 84.1 (78.9–88.5) 87.0 (77.4–93.6) 95.3 (91.8–97.3) 63.8 (56.5–70.5) 35.4 (16.8–75) 84.8 (80.4–88.6)
ALSDI >4 81.6 (76.1–86.3) 89.6 (80.6–95.4) 96.1 (92.7–97.9) 61.1 (54.3–67.4) 38.2 (17.1–85.2) 83.5 (79.0–87.5)
ALSDI >5 66.9 (60.6–72.9) 98.7 (93–100) 99.4 (95.8–99.9) 49.0 (44.5–53.6) 154 (21–>1,000) 74.7 (69.5–79.4)
Test cohort
Awaji 84.8 (73.9–92.5) 72.0 (50.6–87.9) 88.9 (80.9–93.8) 64.3 (49.2–77.0) 14.4 (4.8–43.4) 81.3 (71.8–88.7)
ALSDI ≥4 83.3 (72.1–91.4) 84.0 (63.9–95.5) 93.2 (84.8–97.1) 65.6 (52.0–77.1) 26.3 (7.5–91.3) 83.5 (74.3–90.5)
ALSDI ≥5 71.2 (58.8–81.7) 96.0 (79.7–99.9) 97.9 (87.3–99.7) 55.8 (46.2–65.1) 59.4 (7.5–>400) 78 (68.1–86.0)
Subgroup analysis
Awaji
Positive 82.5 (77.3–86.9) 70.6 (44.0–89.7) 97.7 (95.3–98.9) 21.1 (15.1–28.6) 11.3 (3.8–33.7) 81.8 (76.7–86.1)
Negative 79.2 (65.0–89.5) 91.8 (83.8–96.6) 84.4 (72.5–91.8) 88.6 (81.7–93.2) 42.3 (15.0–119.9) 87.2 (80.3–92.4)
UMN signs
<3 79.4 (67.3–88.5) 92.8 (83.9–97.6) 90.9 (81.0–95.9) 83.1 (75.1–88.9) 49.2 (16.5–147.3) 86.4 (79.3–91.7)
≥3 82.3 (76.8–86.9) 70.6 (44.0–89.7) 97.5 (94.2–98.9) 22.2 (15.9–30.1) 11.1 (3.7–33.3) 81.3 (76.0–85.6)
Abbreviations: ALS = amyotrophic lateral sclerosis; ALSDI = amyotrophic lateral sclerosis diagnostic index; CI = confidence interval; NPV = negative predictive
value; OR = odds ratio; PPV = positive predictive value; UMN = upper motor neuron.
The diagnostic utility of the ALSDI was assessed in the training and test cohorts as well as an ALS subgroup based on Awaji criteria (positive: definite/probable/
possible; negative) and degree of UMN dysfunction (UMN <3: predominant lower motor neuron phenotype; UMN ≥3: prominent UMN signs).
established in the LMN phenotypes, including patients ini- It could also be argued that the control group did not contain
tially classified as Awaji negative. a sufficient number of “difficult” ALS mimicking patients, and
that experienced clinicians would reliably differentiate such
A potential limitation of the ALSDI relates to wider applica- disorders from ALS. It should be stressed, however, that di-
bility, given that the threshold tracking TMS technique agnostic delays remain a major limiting factor in the man-
requires specific technology and expertise. SICI, a major agement of patients with ALS and recruitment into clinical
component of the ALSDI, is derived by utilizing the threshold trials.48 Conditions such as acquired neuromyotonia, Ken-
tracking TMS technique, which is not commercially available nedy disease, Hirayama disease, and inclusion body myositis
at present. Of note, commercialization of the threshold are “difficult” ALS mimics for both frontline clinicians and
tracking TMS technology is in the final stages with the experienced ALS physicians.49 Of importance, these con-
prospect of imminent widespread availability. Once available, ditions comprised 58% of the non-ALS cohort and were re-
multicenter studies incorporating cross-validated study de- liably differentiated from ALS by the ALSDI, underscoring the
sign would enable further assessment of the diagnostic utility diagnostic potential of the novel index.
of the ALSDI in ALS cohorts. It has also been previously
argued that subtle differences in cohort characteristics,26 An ongoing challenge in the management of patients with
such as age, functional status, and disease duration, could ALS pertains to early and definitive diagnosis of ALS.7 Cur-
have reduced the diagnostic utility of the ALSDI. This rent reliance on consensus criteria that are based on identi-
possibility seems unlikely given the robust diagnostic per- fying concomitant UMN and LMN dysfunction, and have not
formance of the ALSDI in both the independent validation been verified in independent cohorts, could result in signifi-
test cohort and in subgroups of clinical interest. Separately, cant diagnostic delays.16,21 Ultimately, delayed recruitment of
none of the patients recruited in the present study un- patients with ALS into clinical trials may occur, perhaps be-
derwent an autopsy and as such the proportion of patients yond the therapeutic window period when therapies may be
with LMN syndrome exhibiting pathologic features of ALS, most effective. The index test (ALSDI ≥4) could enhance ALS
classified as negative on the ALSDI, could not be precisely diagnosis, especially in early stages of the disease process or in
determined. atypical phenotypes, and should be used in conjunction with
Updated Information & including high resolution figures, can be found at:
Services http://n.neurology.org/content/early/2019/02/01/WNL.0000000000006
876.full
Subspecialty Collections This article, along with others on similar topics, appears in the
following collection(s):
Amyotrophic lateral sclerosis
http://n.neurology.org/cgi/collection/amyotrophic_lateral_sclerosis_
Class I
http://n.neurology.org/cgi/collection/class_1
EMG
http://n.neurology.org/cgi/collection/emg
TMS
http://n.neurology.org/cgi/collection/tms
Permissions & Licensing Information about reproducing this article in parts (figures,tables) or in
its entirety can be found online at:
http://www.neurology.org/about/about_the_journal#permissions
Reprints Information about ordering reprints can be found online:
http://n.neurology.org/subscribers/advertise
Neurology ® is the official journal of the American Academy of Neurology. Published continuously since
1951, it is now a weekly with 48 issues per year. Copyright © 2019 American Academy of Neurology. All
rights reserved. Print ISSN: 0028-3878. Online ISSN: 1526-632X.