Part 2 · Diagnosis & course
Diagnosis & criteria
No single test proves ALS. The diagnosis rests on two kinds of motor neuron signs that spread over time, with other causes ruled out. Press Play to follow one composite case through three published criteria sets, then enter findings in the calculator to see how each set classifies them.
Illustrative composite case, not a real person: the timings were chosen to show where the three rule sets differ. Schematic body map, not to scale.
Criteria calculator
The El Escorial criteria were built for clinical trials, not for diagnosing one person, and all three sets are used to classify people for research and trial entry. The calculator shows how each published rule set classifies the findings you enter. It cannot diagnose ALS or rule it out, and it is not medical advice. 136,523
The rules this calculator applies
Gold Coast 2020. All three are required, and the answer is ALS or not ALS, with no grades. R1: progressive motor impairment documented by history or repeated assessment, preceded by normal motor function. R2: UMN and LMN dysfunction in at least 1 body region (in the same region if only one region is involved), or LMN dysfunction in at least 2 regions. R3: investigations exclude other disease processes. UMN signs are clinical only. LMN signs in a muscle are weakness and wasting, or EMG showing both chronic neurogenic change and ongoing denervation (fibrillations, positive sharp waves or fasciculation potentials). A limb region needs 2 muscles supplied by different roots and different nerves; the bulbar and thoracic regions need 1 muscle. 194,463,471,501
Revised El Escorial 2000. Needs LMN evidence, clinical UMN evidence and progressive spread, with other causes excluded. Definite: UMN and LMN signs on clinical evidence in 3 regions. Probable: UMN and LMN signs in at least 2 regions, with some UMN signs rostral to (above) the LMN signs. Probable laboratory-supported: clinical UMN and LMN signs in only 1 region, or UMN signs alone in 1 region, plus LMN signs on EMG in at least 2 limbs. Possible: UMN and LMN signs together in 1 region, UMN signs alone in 2 or more regions, or LMN signs rostral to UMN signs. Under El Escorial EMG rules, denervation meant fibrillations or sharp waves; fasciculation potentials did not count. 75,95,136,184
Awaji 2008. EMG evidence of chronic neurogenic change counts the same as clinical LMN signs in a region, and fasciculation potentials count like fibrillations when chronic neurogenic change is present. The probable laboratory-supported category is deleted, leaving definite, probable and possible with the same region counts. 134,184,518
Where readers differ, this tool follows the quoted text and shows a note. Awaji "probable" is applied as worded (signs in at least 2 regions with some UMN signs rostral), not with the stricter reading that UMN signs must be in 2 regions. Laboratory-supported uses "at least 2 limbs" from the WFN text, not "2 regions". For Gold Coast, UMN and LMN signs in two different regions are counted, because the box only requires one shared region "if only one body region is involved". 95,136,184,248,463
How well do the criteria perform?
An individual-participant meta-analysis of 5 cohorts (Australia, China, Japan, Denmark and Europe, Italy) pooled 3007 people with suspected ALS, 2443 of whom had ALS at follow-up. The bars show pooled sensitivity and specificity with 95% confidence intervals. 471
Certainty for the specificity estimates was rated very low. In borderline cases (leaving out rEEC and Awaji "definite"), Gold Coast sensitivity was 0.95 against 0.82 for both older sets. Leaving out PLS raised rEEC and Awaji specificity to 0.76; Gold Coast stayed at 0.67. 471
| Study | Who was studied | Gold Coast | Awaji | Revised El Escorial | Worth knowing |
|---|---|---|---|---|---|
| Aggregate meta-analysis 2025 520 | 9 studies, 9656 people, all at ALS referral centres | Sensitivity about 95%, specificity 66%, AUC 0.95 | Probable or better: 54% / 94%. Possible or better: 85% / 82% | Probable or better: 59% / 94%. Possible or better: 84% / 77% | How well this applies outside referral centres is uncertain |
| Sydney 2021 219 | 506 consecutive referrals | Sensitivity 92% | 90.3% | 88.6% | "Possible" counted as positive. Gold Coast held up across onset site, duration and disability |
| Five European centres 2021 432 | 404 referrals for EMG: 272 ALS, 94 mimics | Sensitivity 88.2% | Best older set (Awaji with "possible"): 77.6%. Specificity high for all | The gain was "mainly due to" 28 people with PMA | |
| Peking Union 2021 465 | 1185 people, 1162 with ALS | Sensitivity 96.6% | 85.3% | 85.1% | Only 23 people without ALS, so specificity is hard to judge 471 |
| Japan 2024 413 | 639 referrals: 529 ALS, 110 mimics | Sensitivity 96.8%, specificity 77.3% | 89.6% / 65.5% | 89.2% / 66.4% | Gold Coast false positives were mainly cervical spondylosis and parkinsonian syndromes |
| Brescia 2024 182 | 131 ALS, 104 controls, real-world | Sensitivity 93.1% first visit, 96.1% last; specificity 28.8% | 77.8% / 89.3%; specificity 43.3% | 71.8% / 87%; specificity 45.2% | Low specificity for all three in this cohort |
| Calgary 2022 249 | 206 people with ALS | Sensitivity 96.1% | – | Definite or probable: 47.6% | Eligibility for three phase 3 trials rose from 31–45% to 95–100% |
| Dutch and Belgian registers 2022 138 | 5957 people | 600 (10.1%) met Gold Coast only | – | 95 (1.6%) met rEEC only | Decline 0.81 against 0.84 ALSFRS-R points a month; median survival 18.7 against 17.8 months |
| Awaji meta-analysis 2012 119 | 8 studies, 1187 patients | – | Sensitivity 81.1% | 62.2% | Awaji classified 23% (12–33%) more people as probable or definite, more so in bulbar onset |
| Awaji IPD analysis 2016 198 | 1086 patients: 881 ALS, 205 mimics | – | 0.70; updated Awaji 0.73 | 0.58 | The "updated Awaji" restores a laboratory-supported category |
| Meta-analysis 2015 248 | 9 studies, 1121 people with ALS | – | Awaji "not always superior": it moved 20% of rEEC laboratory-supported cases down to possible | A real controversy about Awaji "probable" | |
| Inter-rater study 2019 250 | 399 patients rated blind by 8 physicians | – | Sensitivity 63% | 64% | Agreement was low, worst for "possible" and "probable laboratory-supported" |
Is Gold Coast too generous?
Its higher sensitivity comes partly from counting progressive muscular atrophy (LMN signs only) as ALS. Specificity is lower and varies between cohorts, so mimics still need to be excluded and the diagnosis reviewed. Cervical spondylosis can be misclassified, and several authors of the pooled analysis helped write Gold Coast, a declared possible bias. 413,432,471,520
Trial groups barely changed
In the Dutch and Belgian registers, people classified by Gold Coast progressed and survived much like those classified by revised El Escorial, so the wider net did not make trial populations noticeably more mixed. 138
No more "possible ALS"
The UK MND Clinical Studies Group wrote that the old terms (suspected, possible, probable) implied doubt that was "often psychologically distressing". Under Gold Coast the answer is ALS or not ALS. 501
Categories are not prognosis
In the Irish register, the category at diagnosis did not predict prognosis. Of 165 people first classed as possible or suspected, 67% later became trial-eligible, but 10% of the 254 who died never did. 496
The diagnostic pathway
What guidelines and reviews describe, in the order it usually happens. Real work-ups overlap, and tests follow the presentation.
Suspicion and referral
NICE NG42 lists early isolated symptoms: loss of dexterity, falls or trips, speech or swallowing problems, tongue fasciculations, weakness, wasting, twitching, cramps or stiffness, breathing problems such as breathlessness lying flat or morning headache, and behavioural change or frontotemporal dementia. It recommends referring without delay and naming the possible diagnosis in the letter. EFNS says assessment by an experienced neurologist should happen as soon as possible. 20,401
History and examination
Steadily progressive, often asymmetric weakness without significant sensory symptoms, starting in a leg in about 35%, an arm in about 30% and speech or swallowing in about 30%. Widespread fasciculations, bilateral tongue wasting, the "split hand", head drop and emotional lability have a high positive predictive value (not formally validated). The examination covers all four regions for UMN and LMN signs, and looks for signs against ALS: sensory loss, eye-movement palsy, sphincter disturbance. A three-generation family tree is recommended. 194,502
EMG and nerve conduction
In ALS, motor and sensory conduction is usually normal, with no conduction block. Needle EMG shows fasciculation potentials, fibrillations and positive sharp waves with large, long-duration motor units, and all four regions should be sampled. Conduction block, demyelination, prominent sensory change, or denervation in one nerve territory rather than a myotome point elsewhere. 136,194,502
EMG is not a stand-alone test: in one tertiary series it was "only 60% sensitive", and bulbar-onset ALS can show a normal EMG. 501,502
MRI of brain and spine
Used to exclude lesions of the brain and spinal cord. Some people with ALS show T2 signal along the corticospinal tracts. Brain MRI helps most with isolated swallowing problems or one-sided weakness, cervical MRI with mixed signs and nothing above the neck, and lumbar MRI with an LMN-only leg syndrome. 194,502
Blood tests (EFNS panel)
- electrolytes, urea, creatinine; fasting glucose
- AST, ALT, LDH, albumin; calcium
- TSH, free T4 and free T3
- full blood count; CRP; creatine kinase
- serum protein electrophoresis and immunoelectrophoresis
- vitamin B12 and folate
- Borrelia, HIV, other viruses and syphilis if there is exposure risk
- plus chest X-ray
Controversy These lists apply the panel to everyone; Turner and Talbot argue there are "no mandatory investigations" and that tests should follow the clinical picture to avoid delay. 20,194,202,502
Selected cases
Depending on the picture: GM1, anti-Hu, anti-MAG, AChR and MuSK antibodies, hexosaminidase A/B, ACE, lactate, DNA tests (SOD1, SMN, SBMA, TDP43, FUS), CSF, urine heavy metals, transcranial magnetic stimulation, or muscle or nerve biopsy. CSF matters most when an inflammatory neuropathy, primary progressive MS or a paraneoplastic disorder is suspected. 202,502
In France, everyone with ALS is offered C9orf72 and SOD1 testing at diagnosis, with a wider panel for onset before 40, unusual progression or a family history. See Genes. 194
Neurofilament light (NfL) Early evidence
A protein released by damaged nerve fibres. Serum NfL above 62 pg/mL separated ALS from other conditions with 85.5% sensitivity and 81.8% specificity in one centre. CSF NfL at 2200 pg/mL gave 77% sensitivity and 85% specificity. On average CSF NfL was 3.35 times higher in ALS than in mimics, and serum NfL 8.15 times. 187,484,517
NfL is also raised in stroke, MS and other neurodegenerative diseases, no clinical cut-off is agreed, and adding it did not raise the sensitivity of rEEC or Awaji. It "should not be considered in isolation". 136,269,461,471
Giving the diagnosis, or waiting
If Gold Coast is met with typical UMN and LMN signs, the diagnosis can be given. In early disease, or when two diagnoses remain possible, EMG is repeated at 3 or 6 months depending on how fast things change. 194
Diagnostic delay
Medians and means come from different designs and are not directly comparable. 214,414
The interval from first symptom to diagnosis has stayed at about a year, roughly a third of median survival from onset. Faster disease is referred sooner, so true mimics tend to be slow. 502
In a US centre, 52% of people received another diagnosis first, and they saw 3 physicians on average; delay was longer over 60, in sporadic disease and with limb onset. In six countries, 35% had at least one prior misdiagnosis. Across 13 studies, bulbar onset was diagnosed about 4.4 months sooner than limb onset. 214,414,460
Being told the diagnosis
Who and how (NICE NG42)
A consultant neurologist with up-to-date MND experience gives the diagnosis, unless it is urgent. They ask how much the person wants to know and whom to involve, give oral and written information, provide a single point of contact, offer a face-to-face follow-up with a member of the multidisciplinary team within 4 weeks, inform the GP without delay, and leave time for questions such as "How long will I live?" and "Will my children get MND?" 401
The start of care
EFNS: a consultant who knows the patient and the disease gives the news, followed by regular multidisciplinary support. Breaking the news is described as the start of treatment and an ongoing process: honest while keeping hope, without collusion with relatives, with written material that supports but never replaces a conversation. 20,112
What people said
In an Australian survey of 248 people, 36% were dissatisfied with how they were told. Satisfaction was higher with more time, an empathic response, realistic goals, exploring expectations and making a plan. 24
Only when it is secure
The Scottish register's lesson was to tell people "only when it is secure". The UK MND Clinical Studies Group adds that the clinician should know the proposed care plan. 133,501
A second axis: thinking and behaviour
The ALS-FTSD criteria (2017) describe the motor disease on one axis and cognition and behaviour on another: ALSci (cognitive impairment: executive or language), ALSbi (behavioural impairment: apathy, or at least 2 behavioural features), ALScbi (both), and ALS-FTD. Over 50% of people with ALS have some neuropsychological change, and screening with the ECAS or ALS-CBS is recommended. More in Living with ALS. 194,486