Diagnostic criteria

What do the UK criteria say?

Annex B sets out the following. Definite sporadic CJD requires neuropathological or immunocytochemical confirmation. Probable sporadic CJD requires rapidly progressive dementia plus at least two of myoclonus, visual or cerebellar problems, pyramidal or extrapyramidal features, and akinetic mutism, together with either a typical electroencephalogram showing generalised triphasic periodic complexes at approximately one per second, or the clinical criteria for possible sporadic CJD plus a positive assay for 14-3-3 protein in cerebrospinal fluid. Possible sporadic CJD requires rapidly progressive dementia, two of those four features, and a duration of less than two years.

Definite iatrogenic CJD requires a neuropathological diagnosis of CJD in a patient with a recognised risk factor. Probable iatrogenic CJD is either a progressive predominantly cerebellar syndrome in a recipient of human pituitary growth hormone, or a clinical diagnosis of probable CJD in a patient with a recognised risk factor. The listed exposures are treatment with human pituitary growth hormone, human pituitary gonadotrophin or human dura mater graft; a corneal graft from a donor classified as definite or probable human prion disease; exposure to neurosurgical instruments previously used in a definite or probable case; and transfusion of blood from a donor later diagnosed with variant CJD. The annex states that this list is provisional because previously unrecognised mechanisms may occur.

Definite genetic transmissible spongiform encephalopathy requires neuropathological confirmation plus either definite disease in a first degree relative or a pathogenic PRNP mutation. Probable genetic disease is a progressive neuropsychiatric disorder plus either definite or probable disease in a first degree relative or a pathogenic PRNP mutation. The annex tabulates mutations by neuropathological phenotype, and includes a separate list of variants associated with a neuropsychiatric disorder where prion disease has not been proven.

Definite variant CJD requires a progressive neuropsychiatric disorder with neuropathological confirmation showing spongiform change and extensive PrPSc deposition with florid plaques. Probable variant CJD has two routes. The first requires a progressive neuropsychiatric disorder lasting more than six months with routine investigations not suggesting an alternative diagnosis, at least four of five features (early psychiatric symptoms, persistent painful sensory symptoms, ataxia, myoclonus or chorea or dystonia, and dementia), an EEG without the typical appearance of sporadic CJD or no EEG performed, and symmetrical high signal in the posterior thalamus on MRI, with no history of potential iatrogenic exposure and no evidence of a familial form. The second route substitutes a tonsil biopsy positive for PrPSc. Possible variant CJD drops the imaging and biopsy requirement while keeping the duration, the four of five features and the exclusions.

Annex B also records three categories the Edinburgh unit used for referred patients who do not meet the criteria for possible CJD, the first being cases where the criteria are not met and no reasonable alternative diagnosis has emerged.

What do the CDC criteria say?

Definite sporadic CJD is confirmed by neuropathology or immunodiagnostics, with qualifying methods given as standard neuropathological techniques, immunocytochemistry, Western blot showing protease-resistant PrP, or scrapie-associated fibrils.

Probable sporadic CJD has two routes in the version on the page revised in January 2026. The first is a neuropsychiatric disorder plus a positive RT-QuIC result in CSF or other tissues. The second is rapidly progressive dementia with at least two of myoclonus, visual or cerebellar signs, pyramidal or extrapyramidal signs and akinetic mutism, plus at least one positive supporting test from EEG, 14-3-3 or MRI, and no routine investigations indicating another diagnosis. Possible sporadic CJD is progressive dementia with at least two of the same four features, no positive result on the tests listed for probable, an illness duration under two years, and no routine investigations indicating an alternative diagnosis.

The supporting tests are specified. MRI qualifies on high signal in the caudate or putamen, or in at least two cortical regions from temporal, parietal and occipital, on diffusion-weighted or FLAIR imaging. EEG qualifies on a typical pattern of periodic sharp wave complexes at any point in the illness. A positive CSF 14-3-3 assay counts when disease duration is under two years. The CDC page notes separate criteria for iatrogenic and familial cases.

What did the 2021 international review change?

It proposed amended criteria that incorporate RT-QuIC, so that a positive case can be classified as probable sporadic CJD in early clinical stages. Definite diagnosis still requires neuropathological confirmation. The review states that a diagnosis of possible sporadic CJD, meaning the absence of suggestive biomarkers, should only be made where extensive diagnostics have not revealed alternative explanations.

Its MRI criteria require restricted diffusion in the caudate, in caudate and putamen, in caudate, putamen and thalamus, or in at least two cortical regions from temporal, parietal and occipital, with no subcortical white matter involvement and no isolated restricted diffusion in the thalamus. Diffusion-weighted sequences are required to confirm the pattern.

How well do the component tests perform?

Figures given in the 2021 review, drawn from the studies it covers: CSF RT-QuIC sensitivity 73 to 89 percent for the first generation assay and 92 to 97 percent for the second generation, with specificity of 99 to 100 percent. CSF 14-3-3 sensitivity 92 percent in a 2012 meta-analysis, lower at around 60 to 70 percent in the MV2 and MM2 subtypes, with specificity of 80 percent in that meta-analysis and a reported range of 40 to 92 percent. CSF total tau sensitivity and specificity around 90 percent in most studies, with some reporting 67 percent or below 50 percent, and cut-offs varying between 1,072 and 1,400 pg/mL. MRI on diffusion-weighted or FLAIR imaging gives sensitivity of about 80 percent in some studies and 92 to 98 percent in others, with specificity of 74 to 98 percent. EEG periodic sharp wave complexes give 64 percent sensitivity in older data and 39 to 45 percent in recent comparisons, with specificity of 91 percent.

The US surveillance centre reports its own second generation RT-QuIC at 90.3 percent sensitivity across all forms of prion disease and 98.5 percent specificity, and publishes reference ranges and a scale of post-test likelihoods: a positive RT-QuIC carries over 98 percent likelihood of prion disease, while a negative RT-QuIC with total tau under 499 pg/mL carries under 0.2 percent.

What work-up do the criteria assume?

Clinical investigation, blood sampling and lumbar puncture, with MRI and EEG as a minimum, in the 2021 review. MRI should include T1 with contrast, FLAIR and diffusion-weighted sequences with ADC maps. Where available, RT-QuIC should be performed in every suspected case, 14-3-3 remains the primary surrogate marker, and total tau and the phosphorylated to total tau ratio are supportive. Genetic analysis of PRNP should be considered in all suspected cases. Routine blood, CSF and imaging should rule out common mimics, and brain biopsy is reserved for inconclusive cases where a treatable alternative is suspected.

The UK guidance set also uses the criteria outside diagnosis. Its infection control part classifies patients by risk, with symptomatic patients defined as those who fulfil the criteria for definite, probable or possible CJD, and a separate category for patients with neurological disease of unknown aetiology in whom CJD is being actively considered. That classification is what drives instrument decisions, covered in hospitals and care homes.

Alternative diagnoses are the subject of conditions that can look similar, the referral and notification routes are in suspected case pathway, and the section index is for clinicians.