Abstract
Background: The increase of the 14-3-3 protein in CSF is used as a diagnostic test in Creutzfeldt-Jakob disease (CJD), but the sensitivity and specificity of the 14-3-3 test are disputed. One reason for the dispute may be the recently established heterogeneity of sporadic CJD. The relationship between CSF 14-3-3 protein and sporadic CJD subtypes, distinguished by electrophoretic mobility of proteinase K-resistant prion protein (PrPSc) and genotype at codon 129 of the prion protein gene, has not been elucidated.
Methods: The authors examined the 14-3-3 protein test in 90 patients with sporadic CJD. PrPSc type (type 1 or type 2) and the genotype at polymorphic codon 129 were determined in each patient. Mutations were excluded by prion gene sequencing.
Results: The authors’ findings indicate that the sensitivity of the 14-3-3 test is higher in patients with molecular features of the classic sporadic CJD than in patients with the nonclassic CJD subtypes. The difference appears to be related to the PrPSc type and not to the codon 129 genotype. Disease duration before 14-3-3 testing might also have an influence because it was shorter in classic sporadic CJD.
Conclusion: The Creutzfeldt-Jakob disease clinical subtype should be considered when interpreting results of the 14-3-3 test.
Creutzfeldt-Jakob disease (CJD), the most common human prion disease, is characterized by subacute neurologic deterioration, spongiform degeneration of gray matter, and brain accumulation of proteinase K-resistant prion protein (PrPSc). The majority of CJD cases occur sporadically, with no prion protein gene (PRNP) mutation and no history of exposure to exogenous prions. Recent studies of sporadic CJD (sCJD) from our laboratory have identified specific disease subtypes based on electrophoretic mobility of proteinase K-resistant PrPSc, which is distinct in PrPSc type 1 and type 2, and PRNP genotype at codon 129.1 In the most common sCJD molecular subtype, PrPSc is of type 1, and patients are homozygous methionine at codon 129 (sCJDMM1). This subtype, along with the sCJDMV1 molecular subtype, is associated with the classic CJD clinical phenotype, which is characterized by advanced age at onset and rapid course and a presentation that includes cognitive decline and EEG periodic sharp wave (PSW) discharges. Combined, sCJDMM1 and sCJDMV1 account for ∼70% of all cases of sCJD.2 The other molecular subtypes are associated with less typical CJD clinical phenotypes, which typically have a younger age at onset (sCJDVV1), ataxia at presentation (sCJDMV2 and sCJDVV2), and a slower course and the lack of PSW at EEG examination.
The heterogeneity of the clinical presentation makes the diagnosis of sCJD challenging. Additional confounding factors are the lack of reliable diagnostic tests aside from histopathologic examination of brain tissue. Considerable efforts have been made to identify reliable markers of sCJD using CSF or peripheral tissues. Whereas PrPSc has yet to be consistently detected in CSF, other markers, including 14-3-3 protein, S-100 protein, neuron-specific enolase, and tau protein, have been reported to be useful for the diagnosis of sCJD.3–6⇓⇓⇓ Of these, the CSF 14-3-3 test has shown the most consistent results in sCJD patients, although a number of other neurologic diseases have also been shown to be associated with an increase of the 14-3-3 protein in CSF.7–10⇓⇓⇓
The 14-3-3 protein belongs to a highly conserved family of acidic 30-kDa proteins that play an important role in cell differentiation, proliferation, and signal transduction.11–22⇓⇓⇓⇓⇓⇓⇓⇓⇓⇓⇓ It has also been shown to regulate rate-limiting enzymes in catecholamine and serotonin biosynthesis.23,24⇓ In rat brains, the 14-3-3 proteins, as well as its homologous protein in Drosophila, have been localized to presynaptic boutons, suggesting a role in neurotransmission.25,26⇓
Limited data regarding sensitivity and specificity of 14-3-3 protein in the diagnosis of sCJD are available. The initial description of CSF 14-3-3 in CJD indicated 96% sensitivity and 96% specificity.27 After excluding patients with dementia plus recent stroke, the specificity increased to 99%. However, patient selection based on clinical data, as in the aforementioned study, likely results in an over-representation of the classic phenotype, which comprises only ∼70% of all patients, and in the exclusion of atypical patients.2 Thus, more rigorous characterization of the sCJD patients is indicated, including confirmation of the diagnosis of sCJD and sCJD subtype determination, before drawing firm conclusions about sensitivity and specificity.28 Investigators in Germany have been collecting data in a similar fashion and have recently reported their findings on a limited number of patients.28,29⇓
We have now determined the level of the 14-3-3 protein in the CSF from additional patients with definitely proven and well-characterized sCJD to examine the relationship between CSF 14-3-3 protein and sCJD molecular subtype and to compare the sensitivity of the 14-3-3 protein test in the classic subtype with that in the other subtypes of sCJD.
Methods.
Patient selection.
The 90 patients selected for the study were examined at the National Prion Disease Pathology Surveillance Center (NPDPSC) for an ∼3-year period and had CSF obtained premortem and fixed and unfixed brain tissue obtained at autopsy or biopsy (table 1). Eleven of the 90 subjects were examined initially at the Department of Neurologic Sciences in Bologna, Italy, with subsequent analysis at the NPDPSC. Fifteen of the 90 patients were examined at the NIH after the 14-3-3 protein from several patients and control subjects was comparatively analyzed in both centers to ascertain the constancy of the results. The diagnoses of sCJD subtype were definitely established in all patients by Western blot analysis of PrPSc from brain samples and PRNP sequencing to establish PrPSc type, the genotype at codon 129, and to rule out the presence of pathogenic mutations (see table 2 for sCJD nomenclature and table 3 for molecular-genetic types with corresponding phenotypic features).1,2⇓
Table 1 Cases of sporadic CJD examined shown as total and subdivided in the six molecular subtypes, with the corresponding sensitivity of 14-3-3 protein test
Table 2 Nomenclature and abbreviations used to identify the subtypes of sporadic CJD and corresponding disease phenotypes
Table 3 Molecular subtypes of sporadic CJD with corresponding phenotypic features
Molecular genetic analysis.
The coding region of PRNP was amplified in all patients by PCR as previously described,1,2⇓ and the product was visualized on a 1% agarose gel to detect potential insertion mutations or deletions. Mutations were also ruled out by direct sequencing of PRNP. The genotype at codon 129 was examined in all patients by digestion with the restriction endonuclease NspI or by sequencing of PRNP coding region.
Western blot analysis of PrPSc.
Immunoblot analysis of PrPSc type 1 and type 2 was performed as described using cerebral neocortex alone or together with other brain regions from all cases.1
Western blot analysis of 14-3-3 protein.
The 14-3-3 protein analysis was performed as originally described with minor modifications.27 CSF in 1-mL aliquots was centrifuged at 14,000g at 4 °C for 5 minutes, and 40 μL of supernatant was added to 20 μL of sample buffer and boiled for 10 minutes. After sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) on 12% Tris-glycine Mini-gels (Novex, Wadsworth, OH), protein was transferred to Immobilon P transfer membrane (Millipore, Billerica, MA). Membranes were then incubated with polyclonal antirabbit antibody to the β-isoform of 14-3-3 protein (Santa Cruz Biotechnology, Santa Cruz, CA), followed by secondary antibody (donkey antirabbit immunoglobulin [Ig] G conjugated with horseradish peroxidase; Amersham, Piscataway, NJ), and developed using the enhanced chemiluminescence plus detection system (Amersham). All samples were run in duplicate along with the following controls: positive (detectable 14-3-3 protein from confirmed CJD subjects), negative (undetectable 14-3-3 protein from confirmed non-CJD subjects), and ambiguous (from non-CJD subjects who had trace levels of 14-3-3 protein). The 14-3-3 test was considered positive only when the 14-3-3 immunoreactivity was comparable with that of the positive control.
Statistical analysis.
The χ2 test was used to compare the sensitivity of the 14-3-3 tests in the various subtypes. When the assumption of the χ2 test was violated, Fisher’s exact test was used. Finally, two-sample t-test was used to compare the average durations before 14-3-3 testing and total duration between the classic and nonclassic groups. All p values were based on a two-sided test of significance.
Results.
Genotype.
All patients were classified with sCJD based on the absence of pathogenic mutation after sequencing of the prion protein gene (PRNP) open reading frame, lack of history supporting the diagnosis of iatrogenic CJD, and lack of history and Western blot data that suggest new variant CJD. Genotyping at codon 129 revealed 58 MM patients, 22 VV patients, and 10 MV patients (see table 1).
PrPSc type.
Brain tissue from all 90 patients was positive for proteinase K-resistant PrPSc of either type 1 or type 2 (figure). Fifty-four (60%) of the patients were associated with PrPSc type 1, and the remaining 36 (40%) patients were associated with PrPSc type 2 (see table 1).
Figure. Western blot analysis of proteinase K-resistant prion protein (PrPSc) in brain and 14-3-3 protein in CSF. (Left) Detection of PrPSc in brains of patients with sporadic Creutzfeldt-Jakob disease (sCJD). Brain homogenates were prepared from two sCJD patients with the MM1 and VV2 phenotypes and from a non-CJD control patient. Samples were either untreated (PK−) or treated with proteinase K (PK+), then denatured in sodium dodecyl sulfate (SDS) sample buffer and run on SDS-polyacrylamide gel electrophoresis (PAGE) gel (12%). PrP bands were detected on Western blots using the anti-PrP monoclonal antibody 3F4. The positions of molecular weight markers are indicated on the left. In the PK-untreated samples (Lanes 1, 3, and 5), broad PrP bands were observed in all subjects with a molecular weight between 25 and 35 kDa. After treatment with PK, three bands of PK-resistant PrPSc representing diglycosylated, monoglycosylated, and unglycosylated forms were present in sCJD MM1 (Lane 2) and sCJD VV2 (Lane 4) patients but not in the negative control (Lane 6). The MM1 patient (Lane 2) showed the electrophoretic mobility pattern of PrPSc type 1 that is characterized by the unglycosylated band migrating at 21 kDa, whereas the VV2 patient showed the electrophoretic mobility pattern of PrPSc type 2 that is characterized by the unglycosylated band migrating at 19 kDa.2 (Right) Detection of the 14-3-3 protein in CSF of sCJD patients. CSF samples from the sCJD subjects (MM1 and VV2) and control patient, each run in duplicate, were analyzed for the presence of the 14-3-3 protein on Western blots using a polyclonal antibody against the 14-3-3 protein as described in detail in the Methods section. The 14-3-3 protein migrated at 30 kDa and was detected in the two sCJD patients, MM1 (Lanes 7 and 8) and VV2 (Lanes 9 and 10), but not in the control patient (Lanes 11 and 12).
Fifty-one (56%) patients had the sCJDMM1 or MV1 molecular subtype that is associated with the classic sCJD (see table 1). Twenty-six (29%) had the sCJDVV2 and sCJDMV2 molecular subtypes that are associated with the ataxic phenotype. Ten patients (11%) had sCJDMM2, which is associated with a phenotype characterized by long duration, cortical disease, absence of myoclonus, and severe spongiosis, and three patients (3%) had sCJDVV1 with the “early-onset” phenotype. The sCJDVV1 patients also typically lack PSWs by EEG and have a relatively long disease duration compared with MM1/MV1 patients, justifying their inclusion in the nonclassic category.
14-3-3 protein in CSF.
The 14-3-3 test was positive in 78 of the 90 sCJD patients examined, corresponding to an overall sensitivity of 87% (see figure, left, and table 1). When the 90 patients were subdivided into the six sCJD molecular subgroups according to the combined PrPSc type and genotype at codon 129, the sensitivities of the 14-3-3 test in these groups were MM1, 94% (45/48); MV1, 100% (3/3); MV2, 57% (4/7); VV2, 84% (16/19); MM2, 70% (7/10); and VV1, 100% (3/3; see table 1). When we examined the 14-3-3 test sensitivity as a function of the PrPSc type, we found it higher, 94% (51/54 subjects), in association with type 1 PrPSc than in association with type 2 PrPSc, 75% (27/36 patients; p < 0.01; table 4). Conversely, when we examined the test as a function of the genotype at PRNP codon 129 regardless of the PrPSc type, we found the sensitivity of the 14-3-3 test in the MM patients was 90% (52/58), 86% in VV patients (19/22), and 70% in the MV patients (7/10; p = NS).
Table 4 Sensitivity of the 14-3-3 test according PrPSc type and sporadic CJD phenotype
Because the 14-3-3 protein test is used to establish a clinical diagnosis, it seems appropriate to combine the sCJDMM1 and MV1 molecular subtypes, which are characterized by the classic sCJD clinical phenotype (e.g., advanced age at onset, rapid course, cognitive or visual impairment at presentation, and PSW at EEG), and separate these two subtypes from all the other sCJD molecular subtypes, which usually lack PSW and myoclonus1,2⇓ and are more heterogeneous at presentation (e.g., prominent ataxia or brainstem signs, slowly progressive cognitive decline, and early age at onset). The ataxic subtypes, sCJDVV2 and sCJDMV2, are relatively common among the nonclassic variants, combined accounting for ∼24% of sporadic CJD cases.2 They are included among the nonclassic variants because 1) the absence of PSWs and myoclonus early in disease; 2) the cognitive decline that appears only after months of disease; 3) the often long disease duration; 4) the brainstem/cerebellar presentation that may delay diagnosis; and 5) the distinct neuropathology definitely distinguish them from the classic subtype.1,2⇓ Overall, the sensitivity of the 14-3-3 protein test in classic sCJD was 94% (48/51 patients) compared with 77% (30/39) with nonclassic presentation (p < 0.025; see table 4).
The separation of classic from nonclassic phenotypes is further justified by the different disease duration in these two subgroups of patients examined in our study. The average total disease duration in all our patients was 5.6 ± 5.5 months with average duration before 14-3-3 testing of 3.6 ± 4.0 months (table 5). In contrast, total duration and the average duration before 14-3-3 testing were 3.7 ± 3.3 months and 2.3 ± 2.1 months for classic subjects. Both were different (p = 0.006 and p = 0.001) from the 8.7 ± 7.1-month total duration and 6.2 ± 5.7-month duration before 14-3-3 testing of nonclassic patients (see table 5).
Table 5 Disease durations before 14-3-3 testing and total
In an attempt to answer the question of whether 14-3-3-negative patients differ with respect to clinical signs, we collected clinical data from eight 14-3-3-negative patients for whom detailed clinical records were available (table 6). Although the paucity of 14-3-3-negative patients precludes statistical evaluation, a variety of clinical signs referable to cortical dysfunction (e.g., cognitive decline, visual hallucinations, and speech and language difficulties) were represented, as were signs referable to brainstem (e.g., vertigo) and cerebellum (e.g., ataxia). Two patients were described as having PSWs on EEG, whereas four others had nonspecific EEG changes. Of the two patients with PSWs, one was MM1 (as expected), whereas the other was MM2. Some patients had myoclonus, although startle myoclonus was not specifically sought or reported in any of the 14-3-3-negative patients. A negative 14-3-3 test could not be accounted for by the disease duration before sampling because the range of disease duration before sampling in the 14-3-3-negative patients was between 3 weeks and 11 months. Interestingly, seven of the eight 14-3-3-negative patients were female. Whether this will prove noteworthy will require further study.
Table 6 Data on 14-3-3-negative subjects
Discussion.
We report that the sensitivity of the 14-3-3 protein test, a common but still controversial CSF test used for the clinical diagnosis of CJD, varies in different molecular subtypes of sCJD. We observed a higher sensitivity in patients with the molecular subtypes that correspond to patients with the classic phenotype of sCJD compared with patients with molecular subtypes corresponding to nonclassic phenotypes (94% vs 77%). This was largely a function of PrPSc type 1 vs type 2 because the 14-3-3 test was positive in 94% of sCJD subjects with PrPSc type 1 and 75% in sCJD with type 2 PrPSc, whereas no significant difference in 14-3-3 sensitivities according to the genotype at codon 129 was detected (90% [52/58], 86% [19/22], and 70% [7/10] in the MM, MV, and VV genotypes). Thus, the PrPSc type may have a greater influence on the 14-3-3 protein test than the PRNP codon 129 genotype. This conclusion is also suggested by the finding that the 14-3-3 test sensitivity appears different in the sCJDMM1 patients (94%) and sCJDMM2 patients (70%), who have the same MM 129 genotype but a different PrPSc type, although clearly more MM2 patients are needed to establish significance.
Two studies similar to ours have been previously carried out. The most recent of these two studies28 reported a 96% sensitivity of the 14-3-3 test in patients with the classic molecular profile and a 67% sensitivity in patients with the nonclassic phenotype, although the separation into classic and nonclassic groups, as we have done for purposes of clinical relevance, was not emphasized in this study. Moreover, when these data are combined with those of the present study, the disparity in 14-3-3 sensitivity between classic and nonclassic sCJD subtypes becomes greater and the significance higher. The potential for error when comparing data obtained from different laboratories is acknowledged; nevertheless, the similarity in overall results is noteworthy. Similarly, 83% and 50% sensitivities in sCJDVV2 and sCJD MV2 patients were observed in the aforementioned study, which is comparable with the respective sensitivities for 14-3-3 protein in this study, which were 84% and 57%. Given the small numbers of these sCJD subtypes, no conclusion as to whether MV2 patients have the lowest sensitivity among the sCJD molecular variants can be reached. It should also be noted that statistical comparisons in 14-3-3 test sensitivity among the nonclassic groups are precluded by the small numbers of such patients, even if data are pooled from the available studies. Future studies with more patients should elucidate potential differences in sensitivity among the rare nonclassic subgroups.
Given the rarity of prion diseases and the existence of surveillance centers in several countries, analysis of large numbers of prion disease patients will require comparisons of data obtained in separate laboratories. Of the 90 patients examined in this study, 15 were obtained in collaboration with the NIH, and 11 were obtained from the University of Bologna. Therefore, a number of measures were taken to ensure comparability of data. First, the techniques used, including selection of primary antibodies, were identical at the three centers. Second, 14-3-3 protein assessment of the patients from the NIH were examined in parallel with the NPDPSC. The NIH results and the NPDPSC results were identical. Third, investigators who trained at the NPDPSC (P.P. and S.C.) performed the laboratory testing of the 11 patients from the Department of Neurologic Sciences in Bologna, and the data were analyzed at NPDPSC based on the criteria outlined in this study. We believe these measures ensured maximum comparability between laboratories and overall reliability of the data because it is logistically not possible to perform all testing at a central location. Moreover, good comparability even between studies is suggested by a smaller study of similar parameters showing virtually identical data.28 Because the techniques are the same, the most likely confounding variable in 14-3-3 analysis between studies would be the threshold levels established for positive and negative reactions. In our study, similar to other studies,28 we required positivity in excess of the ambiguous reaction that is encountered in some negative controls to enhance specificity. The high percentage of positive reactions in classic CJD in this study and agreement with other studies suggest that appropriate sensitivity was maintained with this approach.
Although our 14-3-3 data on sensitivity are consistent with the literature when nonclassic subtypes are excluded, one recent study questions the overall sensitivity of 14-3-3 testing for the diagnosis of sCJD.30 In this study, the authors note 53% sensitivity of the 14-3-3 test in 32 sCJD patients. Conversely, the low number of study patients (32 patients, of which 18 were analyzed by Western blot analysis), the fact that testing was performed at seven different centers using a variety of methods and threshold levels for positivity, the fact that sporadic disease and absence of familial cases were not established through genetic testing, and the fact that sCJD variants were not accounted for preclude an in-depth comparison with the present study. Moreover, although altering threshold levels for positivity, such as in sandwich immunochemiluminometric assays, necessarily affects sensitivity, we maintain an unambiguous standard for a positive reaction and still demonstrate excellent sensitivity for the classic sCJD subtype.31 This suggests that sensitivities in the 50 to 60% range may result in part from inclusion of quantitative techniques with arbitrarily high threshold values, in addition to other possible factors such as inclusion of familial cases that are more often 14-3-3 negative or referral bias in favor of nonclassic patients. This further illustrates the point that 14-3-3 protein in CSF is one of a number of data points in a given patient and cannot be interpreted in isolation.
The basis for the different sensitivity of the 14-3-3 test in the different sCJD subtypes cannot be identified. The four sCJD molecular subtypes—MM1, MV1, VV1, and MV2—that have combined sensitivity of 88% (58 positive/66 total patients) have a cerebral cortical histopathology score, obtained combining the semiquantitative score of spongiform degeneration, astrogliosis, and neuronal loss, of 1.7 (0 to 2.5 scale).2 In contrast, in sCJDVV2, in which the 14-3-3 test has a sensitivity of 84%, the histopathology score is 0.7, comparable with that of the four other subtypes. This finding would indicate that the sensitivity of the 14-3-3 test is not necessarily related to the severity of the sCJD histopathology in the cerebral cortex. Similarly, in sCJDMM2, the 14-3-3 sensitivity was 70%, whereas the cerebral cortical histopathology score of this subtype is high (1.95). Conversely, the spongiform degeneration in sCJDMM2 is made of large, often confluent, vacuoles and morphologically is different from the fine spongiform degeneration that is present in all the other subtypes. Whether the 14-3-3 test sensitivity is related to presence and extent of fine spongiform degeneration in the cerebral cortex remains to be determined. Disease duration or rate of disease progression may be another factor affecting the sensitivity of the 14-3-3 test. Overall, nonclassic sCJD subtypes have significantly longer disease duration and may be more difficult to diagnose clinically because they have a more heterogeneous presentation and low frequency of myoclonus and classic EEG changes. For these reasons, the diagnosis of CJD may be made at a longer time interval after the onset than in classic sCJD. In this study, the 14-3-3 protein test was performed significantly later (6.2 ± 5.7 months) in the nonclassic than in the classic sCJD subtypes (2.3 ± 2.1 months). Therefore, long disease duration or slower progression may have a negative effect on the sensitivity of the 14-3-3 test. In the present study, however, the range of disease duration before 14-3-3 sampling from 3 weeks to 11 months suggests that a negative test cannot be accounted for by disease duration. Similarly, the clinical signs in 14-3-3-negative patients did not specifically differ from the 14-3-3-positive patients. Further conclusions regarding the 14-3-3-negative patients are precluded by the small number of such patients and will require continued evaluation.
The results of this study indicate that 14-3-3 testing is more helpful for the diagnosis of patients with classic than nonclassic clinical presentation. This emphasizes the need for proper review of patient history when interpreting the 14-3-3 test. In early studies reporting excellent sensitivity, specificity, and predictive value of the 14-3-3 test, a variety of possible false-positive findings were noted, which appeared to be related to patient selection.32 In this respect, it is noteworthy that the accuracy of the 14-3-3 test appears to decrease as the clinical level of certainty decreases.33 Likewise, CSF 14-3-3 protein detection in unselected dementia patients was found to be unreliable in one study,10 reporting a wide spectrum of vascular, inflammatory, and neurodegenerative diseases in which the 14-3-3 test was positive. Thus, the 14-3-3 test is useful when interpreted in the context of proper clinical evaluation and with the understanding of its limitations.
Our data on the 14-3-3 sensitivity in sCJD subtypes are consistent with this conclusion. Overall, they show that the test is reliable for the diagnosis of the classic sCJD, whereas it may be falsely negative in a significant number of cases with the clinical presentation of the nonclassic sCJD.
Acknowledgments
Supported by the Centers for Disease Control UR8/CCU515004-01, the National Institutes of Health AG-14359-02, and the Charles S. Britton Fund.
The authors thank Dr. Kimbra Kenney and the late Dr. Clarence Gibbs at the Laboratory of Central Nervous System Studies, National Institute of Neurologic Disorders and Stroke, NIH, for providing cases, and Anne Fortino for assistance in manuscript preparation.
Footnotes
-
See also pages 410, 443, and 450
- Received July 2, 2003.
- Accepted May 14, 2004.
References
Disputes & Debates: Rapid online correspondence
REQUIREMENTS
If you are uploading a letter concerning an article:
You must have updated your disclosures within six months: http://submit.neurology.org
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.