Abstract
Objective: To improve diagnostic criteria for sporadic Creutzfeldt–Jakob disease (CJD).
Methods: Pooled data on initial and final diagnostic classification of suspected CJD patients were accumulated, including results of investigations derived from a coordinated multinational study of CJD. Prospective analysis for a comparison of clinical and neuropathologic diagnoses and evaluation of the sensitivity and specificity of EEG and 14-3-3 CSF immunoassay were conducted.
Results: Data on 1,003 patients with suspected CJD were collected using a standard questionnaire. After follow-up was carried out, complete clinical data and neuropathologic diagnoses were available in 805 cases. In these patients, the sensitivity of the detection of periodic sharp wave complexes in the EEG was 66%, with a specificity of 74%. The detection of 14-3-3 proteins in the CSF correlated with the clinical diagnosis in 94% (sensitivity). The specificity (84%) was higher than that of EEG. A combination of both investigations further increased the sensitivity but decreased the specificity.
Conclusions: Incorporation of CSF 14-3-3 analysis in the diagnostic criteria for CJD significantly increases the sensitivity of case definition. Amended diagnostic criteria for CJD are proposed.
The initial diagnosis of Creutzfeldt–Jakob disease (CJD) is often based on clinical grounds and is usually correct in patients with a progressive dementia extant for < 2 years, with at least two of the following neurologic signs: myoclonus, visual or cerebellar signs, pyramidal or extrapyramidal signs, and akinetic mutism.1 Investigations such as EEG,2,3 ⇓ MRI,4,5 ⇓ and detection of 14-3-3 proteins in the CSF6,7 ⇓ support or weaken the clinical suspicion of the disease, even if definite diagnosis of CJD still requires postmortem examination of the brain.8,9 ⇓ There is an urgent need for empirical data clarifying the diagnostic utility of the combined results of the 14-3-3 test in the CSF and periodic sharp and slow wave complexes (PSWC) in the EEG for confirming CJD. We analyzed the value of these two widely used diagnostic tests in CJD (i.e., EEG and CSF detection of 14-3-3 proteins) in the context of a prospective multinational study of CJD.10 The results of these investigations were used in the classification of suspected CJD cases. Follow-up was obtained, and the accuracy of the initial clinical diagnosis based on established criteria2 was evaluated. Finally, the sensitivity and specificity of new extended diagnostic criteria were analyzed, after inclusion of 14-3-3 CSF results.
Material and methods.
Patients.
Patients were recruited from national registries of CJD in Australia, Austria, France, Germany, Italy, Spain, and the United Kingdom.10 They were classified as definite, probable, or possible CJD cases2 based on available clinical and laboratory data. A standard questionnaire was used to collect data from each national center.
Diagnostic criteria for sporadic CJD.
We used previously established criteria to determine the level of diagnostic accuracy in relation to sporadic CJD.2,10 ⇓ All patients with spongiform encephalopathy or immunochemical detection of PrPSc in the brain were classified as definite CJD cases. Patients were classified as probable cases if they exhibited a rapidly progressive dementia, PSWC on EEG, and two of the following: myoclonus, visual and cerebellar symptoms or both, pyramidal and extrapyramidal signs or both, or akinetic mutism. Those fulfilling the above clinical criteria but without PSWC and a duration of dementia of < 2 years were classified as “possible CJD.” Cases that did not fulfill these clinical criteria were classified as “not CJD.” EEG was used in the initial and follow-up classification of suspected cases. In contrast, results of 14-3-3 determination in CSF were used neither in the initial nor in the follow-up classification of the suspected cases. During follow-up, patients were regularly reviewed and reclassified according to these criteria. Cases with unambiguous clinical improvement or recovery were classified as “other” even if a definite alternative diagnosis could not be made and postmortem analysis was not undertaken.
Analysis of CSF for 14-3-3.
CSF samples were obtained by lumbar puncture. CSF was analyzed for the presence of 14-3-3 proteins as previously described.6,11 ⇓ The CSF analysis was carried out in each CJD surveillance center with minor modifications. Blood-stained CSF samples were excluded from analysis because contamination of this type can lead to “false-positive” results. A blinded comparison of test results from the different centers showed good reliability and validity. (The detailed analysis of the comparison will be published elsewhere.)
Results.
Patient characteristics.
Assessment of CSF 14-3-3 proteins was performed on 1,003 patients who were referred to the national CJD registries of Australia (n = 71), Austria (n = 31), France (n = 270), Germany (n = 351), Italy (n = 79), Spain (n = 79), and the United Kingdom (n = 122). The classification of these patients is reported in table 1. CSF samples of cases classified as “not CJD” with other neurologic disorders served as controls.
Occurrence of 14-3-3 proteins in CSF of Creutzfeldt–Jakob disease (CJD) patients and of patients with other neurologic diseases (n = 1,003)
The 14-3-3 proteins were detected in 91% of patients with definite or probable sporadic CJD, 45% of new variant CJD (nvCJD) and in 97% of CJD patients with a PRNP mutation (see table 1). The test was positive in only two of five Gerstmann–Sträussler–Scheinker (GSS) patients and was negative in all patients with fatal familial insomnia (FFI).
Among control subjects, the test was positive in only ∼ 9%.
Follow-up.
Complete clinical follow-up data were available in 805 cases (sporadic CJD and control subjects). In the remainder, no follow-up was possible because of study design limitations in particular countries. The mean follow-up time varied from 1.8 to 6.5 months (median) between the countries, depending on the study design (range 0.25 to 48 months).
The initial clinical classification of suspected CJD patients at the time of notification to national registries was as follows (table 2). Three hundred thirteen patients were classified as probable CJD cases; among these, 282 had detectable levels of 14-3-3 in the CSF. Two hundred eleven patients had a diagnosis of possible CJD, and CSF was positive for 14-3-3 in 151 of them. Two hundred fifty-nine of 281 patients who did not fulfill the criteria for sporadic CJD (“not CJD”) had no measurable level of 14-3-3 in the CSF.
Initial and final classification in patients with suspected sporadic Creutzfeldt-Jakob disease (CJD)
A further analysis based on the results of follow-up investigations was carried out for each initial diagnostic classification subgroup (see below).
Follow-up on probable CJD cases.
The diagnostic reliability of “probable CJD” was high. During follow-up, the diagnosis of CJD was excluded in only 11 patients; about half of the remaining were confirmed postmortem. In the remaining patients, there was no change in the diagnostic classification because either no autopsy of the brain was performed or the patients were still alive. Of the 11 patients with a false-positive EEG, 14-3-3 proteins were detectable in the CSF of 3. In one of these, however, there were high CSF protein levels and hypercalcemia, leading to a diagnosis of multiple myeloma. In four patients with AD, the diagnosis was made only after autopsy.
Follow-up on possible CJD cases.
The analysis of this subgroup is of particular practical importance, as a significant proportion of sporadic CJD patients never develop typical PSWC. Ninety-three percent of patients in whom postmortem examination of the brain confirmed CJD or who developed typical PSWC during follow-up (and were therefore upgraded to “probable CJD”) had detectable levels of 14-3-3 in the CSF. The 14-3-3 proteins were also detected in 73% of patients whose diagnosis of “possible CJD” did not change during the follow-up, either because autopsy was not done or because the patients were still alive. The 14-3-3 proteins were detected in only 13% of patients in whom the diagnosis of CJD was ruled out.
Follow-up on “not CJD” cases.
Only a small number of patients who were initially classified as “not CJD” at the time of notification turned out to be sporadic CJD cases. Interestingly, five of seven of them had positive results for 14-3-3.
Comparative diagnostic value of EEG and CSF findings.
For the calculation of sensitivity relating to EEG and 14-3-3 results or both, we used CJD patients who were clinically diagnosed as “probable” or “possible” cases and were subsequently confirmed at neuropathologic examination. Accordingly, for the evaluation of specificity, we used only “probable” or “possible” CJD patients who were ultimately neuropathologically excluded or who were clinically recovered (table 3).
Comparative diagnostic value of EEG and CSF analysis in cases of Creutzfeldt–Jakob disease (CJD) initially classified as “probable” or “possible” CJD
A positive 14-3-3 test was more sensitive and more specific than the presence of PSWC in the EEG. The 14-3-3 test gave a false-positive result in only 7 of 43 cases (1 each of anoxia, stroke, metabolic encephalopathy, multiple myeloma, AD, cerebral ischemia with frequent seizures, and 1 patient with disease duration of > 5 years without definite diagnosis), whereas PSWC were seen in 11 of 43 “not CJD” patients (1 each of multiple myeloma and cerebral ischemia and frequent seizures, 4 patients with AD, and 5 with no definite diagnosis neuropathologically).
In patients with PSWC in the EEG and detectable 14-3-3 levels in the CSF, the clinical diagnosis of CJD was confirmed at autopsy in all but three cases. The diagnoses in these three patients were multiple myeloma, AD, and cerebral ischemia with frequent seizures. In contrast, 28 of 60 patients with a clinical diagnosis of “possible CJD” (no PSWC in the EEG), who were also negative for 14-3-3, were ultimately classified “not CJD” (table 2).
The sensitivity of the clinical diagnosis of CJD increases from 66 to 97% with the inclusion of a positive 14-3-3 result in the diagnostic criteria (table 4). Although this procedure appears associated with decreased specificity (from 74 to 65%), it is noteworthy that the majority of false-positive results were due to the presence of PSWC in the EEG. The positive predictive value of PSWC in the EEG was comparable with the detection of 14-3-3 in CSF. The highest positive predictive value was achieved when both the EEG and the 14-3-3 proteins were positive: Ninety-eight percent of these cases were confirmed by neuropathology. The highest predictive value for the absence of CJD was achieved when both investigations were negative (negative predictive value 79%).
Modified diagnostic criteria for sporadic Creutzfeldt-Jakob disease (CJD)
Discussion.
The molecular basis for clinical and neuropathologic phenotypic variability in CJD has recently been described.12 Cases of sporadic CJD with short clinical duration and typical PSWC in the EEG were identified as a distinct subgroup. However, the EEG was rarely typical in other CJD subtypes, leading to the need for improved diagnostic criteria in these forms of the disease. Detection of the pathologic isoform PrPSc outside the CNS has not yet been possible in sporadic CJD in contrast to nvCJD, in which PrPSc has been detected in lymphatic tissues.13,14 ⇓ Although the postmortem rate in CJD is relatively high, between 43 and 85%,10 in a significant proportion of cases, the diagnosis of CJD still depends on clinical criteria. Such criteria need to have a high sensitivity and specificity to guarantee an accurate surveillance of the disease and to ensure a high level of comparability and reliability of epidemiologic data among different countries.
Several investigations have been proposed to corroborate the clinical diagnosis of CJD. In the criteria established by Masters et al.,2 the detection of PSWC in the EEG was used to support the clinical diagnosis of CJD in patients with rapidly progressive dementia and various neurologic signs. Increased levels of neuronal or astrocytic protein have been investigated in the CSF of CJD patients, namely, 14-3-3 proteins,6,7,11,15-17 ⇓ ⇓ ⇓ ⇓ ⇓ neuron-specific enolase, and S-100.15,18-20 ⇓ ⇓ ⇓ Brain CT does not show specific patterns in CJD, but brain MRI may be diagnostically useful in sporadic CJD and nvCJD.4,5,21 ⇓ ⇓ Other methods have been proposed, including somatosensory evoked potentials, retinography, SPECT, and PET,22 but these methods have not been adequately evaluated in large numbers of cases.
Data on the sensitivity and specificity of PSWC in the EEG and 14-3-3 proteins in CSF have been reported in retrospective studies, but they were based on cases that fulfilled established clinical diagnostic criteria or were confirmed neuropathologically.6,7 ⇓ However, the sensitivity and specificity of a test are best evaluated in a prospective study.
The present prospective study has demonstrated that the detection of 14-3-3 proteins in the CSF is sensitive and specific for the diagnosis of sporadic CJD. In familial CJD patients bearing codon 200 or codon 210 mutations, the 14-3-3 assay had a similar diagnostic value. In contrast, 14-3-3 proteins were not detected at all in FFI and only uncommonly in GSS. Both these conditions may be difficult to diagnose clinically because of an atypical clinical picture, often with a long disease duration and absence of PSWC.23 Nevertheless, CSF analysis for 14-3-3 proteins is not useful as a screening test in patients with dementia, as the protein is only a nonspecific marker of neuronal damage. However, in the context of suspected CJD, its presence helps to corroborate the diagnosis and to differentiate these cases from cases of AD and other conditions. False-positive findings may occur in patients with cerebral hypoxia, intracerebral hemorrhage, inflammatory CNS disorders, and blood-stained CSF samples. These conditions can and should be excluded by clinical evaluation and appropriate investigations such as CSF analysis and brain imaging.
Follow-up analysis in the “probable CJD” group demonstrated the high reliability of this clinical classification. Similar analysis of “possible CJD” cases allowed three conclusions: 1) The detection of 14-3-3 proteins strongly supports the diagnosis of CJD in this particular subgroup. 2) The elevation of 14-3-3 proteins in CSF may precede the detection of PSWC in the EEG. 3) Negative CSF 14-3-3 and no PSWC in the EEG strongly support an alternative diagnosis. The analysis in cases who do not fulfill the diagnostic criteria for CJD but who are 14-3-3 positive revealed that careful follow-up is required as this test may detect disease at a very early stage.
Overall, our results demonstrate a high accuracy of the clinical diagnosis of CJD. In our analysis, the 14-3-3 test was positive in only 34 of 392 control subjects, when trying to exclude CJD in patients with a relatively low clinical likelihood. The utility of the 14-3-3 test is further enhanced when interpreted in the clinical context, as many of these “false-positive” findings were due to diseases that do not form a differential diagnosis for CJD, such as subarachnoid hemorrhage or inflammatory disease. In the subgroup of our patients with detailed information on follow-up, 297 of 313 patients with a final diagnosis of “not CJD” were negative for 14-3-3 in the CSF, giving a high specificity of 94%.
The inclusion of a positive 14-3-3 CSF assay in the diagnostic criteria leads to a higher sensitivity. In the framework of this large surveillance study, only a few definite CJD cases did not show either a characteristic EEG or positive CSF for 14-3-3 proteins. Moreover, 14-3-3 proteins were detectable in some patients at an early stage of the disease, preceding the typical EEG changes, suggesting that they may be an earlier disease marker than the EEG. Of importance, when both the EEG and CSF were positive, the highest positive predictive value was achieved. In 98% of these cases, the clinical diagnosis was confirmed by neuropathology (positive predictive value; table 3). The highest predictive value for absence of CJD was achieved when both investigations were negative (negative predictive value 79%), suggesting that in routine clinical practice, alternative diagnoses should be considered in such patients.
Although the results of this study demonstrate worthwhile diagnostic accuracy in patients with “probable” CJD using established diagnostic criteria, the sensitivity of the category “probable CJD” is increased by the inclusion of 14-3-3 assay results. Further, this study also shows that a significant proportion of cases classified as “possible CJD” according to established criteria (absence of PSWC on the EEG) do, in fact, suffer from CJD. As cases of “possible CJD” are often excluded in studies of CJD because of concerns of diagnostic validity, there is a particular need to improve the diagnostic accuracy in this subgroup of cases. It is clearly this group in which the 14-3-3 CSF test appears to be of most practical benefit in classifying patients because without autopsy, CJD may not be diagnosed in these cases. We propose new diagnostic criteria for the classification of CJD cases.
Acknowledgments
Acknowledgment
The authors thank all physicians in Germany notifying suspect cases to the German CJD surveillance unit and for their help in obtaining CSF specimens. Special thanks are given to Maja Schneider-Dominco for data management and assistance in manuscript preparation. The authors thank A. Green, E. Thompson, P. Brown, and M. Harrington of the U.K. study for carrying out the 14-3-3 analyses. They also thank Drs. S. Bevivino and Q.G. Liu for analysis of samples and Dr. P. Cortelli for CSF samples from FFI patients in the study in Italy. Also acknowledged is Ms. Clemencia Plitt for assistance at the CJD National Registry and Drs. Albert Saiz and Justo Carcia de Yebenes for support on laboratory work at Barcelona and Madrid in the study in Spain. In the Austrian study on 14-3-3 proteins, the authors thank Dr. Brigitte Gatterbauer for valuable help and DI Dita Drobna for excellent technical assistance. They are also grateful to Dr. Christa Jarius for diagnostic evaluation of the patients.
Footnotes
This research was funded through the European Commission (BMH4-98-3727). The German study was supported by a grant from the Federal Ministry of Health (Bundesministerium für Gesundheit, GZ 325-4471-02/15). The study in the United Kingdom was funded by the Department of Health and the Scottish Office Department of Health. The study in Italy was partially supported by the National Registry of Creutzfeldt–Jakob Disease, Istituto Superiore di Sanità, financed by the Italian Ministry of Health and by the Department of Social Services of the Italian Ministry of Health. The study in Spain was supported by grants 95/1835 and 99/1332 from Fondo de Inestigaciones Sanitarias. The Austrian studies on 14-3-3 proteins were supported by the Federal Ministry of Science and Traffic (project GZ 70.026/2-Pr/4/98). The Austrian Reference Centre for Human Transmissible Spongiform Encephalopathies was funded by the Federal Ministry of Labour, Health, and Social Affairs. The Australian Creutzfeldt–Jakob Disease Registry was funded by the Commonwealth Department of Health and Family Services.
- Received September 7, 1999.
- Accepted in final form June 6, 2000.
References
Letters: Rapid online correspondence
REQUIREMENTS
You must ensure that your Disclosures have been updated within the previous six months. Please go to our Submission Site to add or update your Disclosure information.
Your co-authors must send a completed Publishing Agreement Form to Neurology Staff (not necessary for the lead/corresponding author as the form below will suffice) before you upload your comment.
If you are responding to a comment that was written about an article you originally authored:
You (and co-authors) do not need to fill out forms or check disclosures as author forms are still valid
and apply to letter.
Submission specifications:
- Submissions must be < 200 words with < 5 references. Reference 1 must be the article on which you are commenting.
- Submissions should not have more than 5 authors. (Exception: original author replies can include all original authors of the article)
- Submit only on articles published within 6 months of issue date.
- Do not be redundant. Read any comments already posted on the article prior to submission.
- Submitted comments are subject to editing and editor review prior to posting.
