A hospital-based registry of Creutzfeldt–Jakob disease: Can neuroimaging serve as a surrogate biomarker?
Correspondence Address: Source of Support: None, Conflict of Interest: None DOI: 10.4103/0028-3886.181538
Source of Support: None, Conflict of Interest: None
Aim: This study addresses the role of neuroimaging in addition to the available clinical criteria for Creutzfeldt–Jakob disease (CJD) and its impact on its diagnosis in the absence of cerebrospinal fluid (CSF) biomarkers and tissue-based approaches.
Keywords: Biomarkers; clinical criteria; Creutzfeldt–Jakob disease; dementia; diffusion magnetic resonance imaging; neuroimaging
Creutzfeldt–Jakob disease (CJD) is a rare yet well-known cause of rapidly progressive dementia (RPD) with a fatal outcome. There are different subtypes of CJD and the estimated worldwide incidence of sporadic CJD (sCJD), the most common subtype of the disease, is around 1–1.5/million. sCJD, being a rare disorder with variably sensitive and specific diagnostic markers, remains a challenge for the treating neurologists. Despite the growing awareness, the condition still remains largely underdiagnosed, especially in resource-poor countries where supportive tests such as cerebrospinal fluid (CSF) 14-3-3 testing  and genetic studies for prion protein mutations  are not readily available and/or are inaccessible due to the prohibitive costs involved. Moreover, the tissue diagnosis is practically not feasible in all cases even in developed countries, given the complexity involved in tissue handling and processing.
Previously, correlation was demonstrated between diffusion signal abnormalities on magnetic resonance imaging (MRI) and higher degree of spongiform changes in the prion infested brains suggesting the potential for neuroimaging to serve as a reliable biomarker for CJD. Research has also demonstrated that the MRI pattern of neocortical and striato-thalamic involvement has a high sensitivity and specificity for diagnosing CJD., This has recently been modified and validated using nonprion RPD controls in a study applying the proposed University of California, San Francisco (UCSF) MRI criteria. Various diagnostic criteria have been proposed for the diagnosis of CJD over the last two decades including the World Health Organization (WHO) criteria, the UCSF criteria, and the European MRI-CJD consortium criteria. However, given the variations in the positive and negative predictive values of these criteria, the diagnosis of CJD still remains difficult. Hence, this research was aimed at studying the clinico-radiological heterogeneity of suspected CJD and the influence of neuroimaging criteria on the diagnosis of 'probable' and 'possible' CJD in a cohort of patients with RPD [see Appendix 1 to view the criteria].
The study group consisted of consecutive patients from January 2000 to December 2013, who were admitted to the inpatient services of a tertiary care hospital in the Southern State of Kerala, India with RPD, where the diagnosis of CJD was suspected. Dementia was diagnosed by the Diagnostic and Statistical Manual (DSM)-IV criteria, and RPD was detected based on a subacute time course and an accelerated rate of decline that developed in fewer than 2 years. Computerized records were used to identify patients with RPD and finally case records were reviewed for further clarification. The demographic, neuropsychological, clinical and imaging characteristics, and the details of biochemical investigations of these patients were reviewed.
All patients underwent evaluation at least once during the course of the illness with an electroencephalogram (EEG) and a 1.5 Tesla MRI scan of the brain including T1- and T2-weighted images, gadolinium-enhanced T1-weighted images, diffusion-weighted images (DWI), fluid attenuation inversion recovery (FLAIR) images, gradient echo sequences and when available, apparent diffusion coefficient (ADC) maps corresponding to DWI images. Diffusion restriction and DWI/FLAIR mismatch was studied in detail to determine if the hyperintensities were real or were artifacts. EEG was recorded using the international 10–20 system. Periodic sharp wave complexes (PSWCs) were defined based on the criteria elaborated by Steinhoff et al.
According to the WHO criteria for CJD, patients were classified as being in the 'probable' and 'possible' CJD groups. None of the patients underwent an autopsy. The other criteria for CJD, namely, the UCSF criteria and the European MRI-CJD consortium criteria were applied to the study cohort to compare the variability in the diagnosis of CJD according to these criteria. In addition, the proposed UCSF MRI criteria were included to focus purely on the radiological components of diagnosis [Appendix 1]. The approval of the Institutional Ethics Committee was obtained for the study. The statistical analysis primarily involved descriptive statistics, and the nonparametric Mann–Whitney U-test was done to compare the three MRI categories of the proposed criteria from the syndromic and electrophysiological perspective.[Additional file 1]
Of the 392 consecutive patients satisfying the DSM-IV criteria for dementia, who were admitted to the inpatient services from January 2000 to December 2013, 96 (24.4%) patients had a diagnosis of RPD. Forty-one (42.7%) patients were clinically diagnosed as having CJD in this cohort according to the WHO criteria, of which thirty (73.2%) patients were diagnosed as 'probable' and 11 (26.8%) patients as 'possible' CJD. The break-up of patients with RPD identified in the cohort is represented in [Figure 1].
The demographic profile of the group satisfying the WHO criteria for CJD is summarized in [Table 1].
Clinical characteristics of 'probable' and 'possible' Creutzfeldt–Jakob disease
All the 41 patients had presented with RPD. The additional features noted were myoclonus (n = 29, 70.7%), cerebellar dysfunction (n = 16, 39%), visual disturbances (n = 11, 26.8%), pyramidal dysfunction (n = 2, 0.5%), extrapyramidal dysfunction (n = 30, 73%), and akinetic mutism (n = 19, 46.3%) [Figure 1]. According to the syndrome of presentation, the patients could be categorized [Figure 2] into mainly one or many of the following overlapping groups which included RPD (n = 41, 100%), seizures including epilepsia partialis continua (n = 5, 12%), psychiatric symptoms (n = 13, 31.7%), akinetic rigid syndrome (n = 15, 36.6%), ataxia (n = 16, 39%), or focal cortical syndromes (n = 24, 58.5%), including aphasia (n = 22), apraxia (n = 1), cortical blindness (n = 2), and neglect (n = 3). In most of the patients, a detailed neuropsychological evaluation could not be done due to the profound cognitive dysfunction. In 12 patients who could be tested, fronto-temporo-parietal dysfunction was notable on neuropsychology testing in the early stages of the disease.
EEG was done in all patients at least once, and of the 41 patients, 28 (68%) had PSWCs suggestive of CJD. Of these, 17 patients had generalized periodic epileptiform discharges (GPEDs), four had periodic lateralized epileptiform discharges (PLEDs), one patient had bilateral independent PLEDs (BIPLEDs), five had coexisting GPEDs and PLEDs, and one had GPEDs along with BIPLEDs. In the remaining 13 patients, there was generalized (n = 12) or focal slowing (n = 1). Nine of the 28 patients had shown focal PSWCs, of whom two patients had an EEG with PSWCs over the posterior head regions. In six out of nine patients, PSWCs were seen with a second, and in four out of five patients, with a third EEG, which were obtained at a median interval of 7 days from the initial EEG. Two patients in the nonprion group (n = 55) had short interval PSWCs during their presentation as encephalopathy and were diagnosed to be having voltage-gated potassium channel antibody-mediated limbic encephalitis and beclouded Alzheimer's dementia with metabolic encephalopathy. Neither of these patients met the MRI criteria for CJD.
The MRI of the brain according to the dementia protocol was done in all patients, of whom thirty (73%) patients had MRI findings suggestive of CJD as per the clinical criteria. They were further categorized into the classes of the proposed UCSF MRI criteria (Vitali et al., 2011). None of the patients suffering from the nonprion RPD satisfied any of the MRI components of each criterion for diagnosis. One of the patients had bilateral DWI hyperintensities involving the pulvinar bilaterally, suggestive of the new variant CJD. Results of the radiological analysis by MRI criteria alone are depicted in [Table 2], which also gives the corresponding EEG characteristics. [Figure 3] depicts the radiological and electrophysiological diagnosis of CJD applying the current criteria.
As shown in [Table 2], PSWCs were seen in 54.5% patients with 'MRI definite CJD,' 87.5% patients with 'probable CJD,' and 72.7% patients with 'MRI not satisfying criteria' for CJD and the prevalence of PSWC between the three groups was not statistically significant (P = 0.241). There were no statistically significant differences between the three MRI categories and the various neurological symptoms mentioned previously with the exception of visual symptoms (hallucinations/visual impairment), which were prevalent in 80% of patients with 'MRI probable CJD' as opposed to 10% of patients with 'MRI definite CJD' and 27.3% with 'MRI not satisfying CJD' criteria (P = 0.005).
Comparing the available criteria for diagnosis of Creutzfeldt–Jakob disease
The diagnosis of CJD as per the currently available criteria in comparison to the proposed MRI criteria (2011) is depicted in [Figure 4].
Thirty-eight patients (92.7%) satisfied both the UCSF 2007 criteria and European MRI-CJD consortium criteria. Thirty patients (73%) satisfied the UCSF MRI criteria, which did not incorporate clinical information, of whom 22 patients were classified as 'MRI definitely CJD' based on the involvement of cingulate, striatum and more than one neocortical gyrus; and, eight patients were classified as 'MRI probably CJD' based on their variable striatal and gyral involvement. CSF 14-3-3 was not done in any of the patients, and hence, in our study group, patients satisfying the UCSF 2007 criteria and European criteria were the identical thirty patients who were confirming to the MRI criteria and the WHO criteria for 'probable CJD.' Eight patients could be diagnosed only when EEG was utilized additionally in accordance with the UCSF 2007 and the European criteria. Three patients who were diagnosed as 'possible CJD' did not meet any of the other criteria and had no autopsy data to suggest etiology. The MRI negative subgroup of 11 patients met the WHO criteria for 'possible CJD.' [Table 3] reflects the combined utility of MRI and EEG criteria in the diagnosis of CJD in our cohort.
The follow-up data was available in 37 patients of which thirty patients (81%) had expired. The mean duration from the onset of symptoms to death was 7.2 months (standard deviation = 4.72; range: 1–18 months). Seven patients were alive at the time of follow-up after 6–12 months and were severely disabled and bed bound. The overall mean survival period till the point of last contact was 9.8 months.
The utility of the WHO criteria as a screening tool for CJD, independent of biomarkers, holds relevance in the subcontinent either due to the nonavailability or absence of in-house logistic capabilities for CSF biomarker assays. All 41 patients with 'probable' and 'possible' CJD presented with RPD along with two or more of the requisite clinical features. Although MRI is not required in the WHO criteria, 73% patients had features on MRI favoring the presence of CJD. Three patients who were diagnosed to be having 'possible CJD' as per the WHO criteria did not meet any of the current criteria for CJD and lacked an autopsy confirmation of the etiology. The clinical inclusiveness of the WHO criteria affords us the unique opportunity to look into the recent criteria. The latter have included the radiological features and CSF biomarkers to compensate for the propensity of the WHO criteria to overdiagnose CJD by considering a group of patients as 'possible' CJD” in whom the clinical changes may potentially be due to an alternative pathology. The demographic and clinical profiles of patients in our series were in concordance with most of the international case series ,, and the available single-center case series from India., The mortality results were similar to that seen in the previously reported data.,
Twenty-eight patients (68%) had the classical EEG changes of CJD in the form of PSWC [Figure 2] similar to the findings of Steinhoff et al (2004). Among the patients with the nonprion RPDs in our cohort, <4% of patients demonstrated the typical EEG features of CJD and these did not meet the clinical or MRI criteria. Notably, ten out of the 13 (77%) patients with suspected CJD who were EEG negative had MRI abnormalities suggesting the presence of a 'definite' or 'probable' CJD indicating that a diagnosis of CJD could not have been made in those ten patients without the help of an MRI. It was previously suggested that the EEG and CSF biomarkers have a limited utility in diagnosing CJD. Our results, along with the fact that nearly one-fifth of the cohort were MRI negative but EEG positive, suggest the independent roles of EEG and neuroimaging in the diagnosis of CJD. The presence of PSWC on EEG was independent of any of the MRI findings according to the proposed UCSF criteria [Table 2]. In patients in our cohort in whom the EEG was not suggestive of CJD, it is possible that repetition of further EEGs could have increased the yield of PSWC.
We assessed the impact of advanced neuroimaging focusing on the role of DWI-ADC and FLAIR signals in the diagnosis of CJD. A recent meta-analysis revealed DWI restriction over the cortex in 80.6% of patients with CJD, basal ganglia in 69.4% patients, thalamus in 30.4% patients, and brain stem/cerebellum in <10% patients. This favored the utility of MRI as a diagnostic tool in CJD, even in the early cases. The stringent radiological criteria with a reported sensitivity and specificity of 96% and 93% respectively, negated the diagnosis of CJD in nearly 27% of our original cohort suggesting the need for confirmation with additional biomarkers. The inclusion of MRI in the European criteria increased its yield in the diagnosis of CJD even in the absence of CSF 14-3-3. This finding was replicated in our study, considering that more than 90% of patients were diagnosed as CJD in our cohort using this criterion. The possibility of overdiagnosis with the European criteria cannot be excluded, as it required an additional support of only one biomarker (EEG, MRI, or CSF), which is an undesirable feature in a devastating, degenerative illness. The key MRI components in the European MRI-CJD consortium criteria  and the UCSF 2007 criteria  which were adapted from the study by Young et al., are almost similar to the definitive criteria in the proposed UCSF 2011 modification in terms of the areas of involvement. The latter, however, emphasized that DWI lesions along with ADC correlates were more significant criteria than the FLAIR hyperintensities in isolation. Categorizing MRI results as 'definite' and 'probable' expresses the level of confidence on which a conclusion can be drawn relying on radiological biomarkers. The proposed modified UCSF MRI criteria will definitely prioritize the need for further investigations to confirm the diagnosis as only 53.7% of patients met the criteria for 'MRI definite' CJD and 19.5% for 'MRI probable' CJD.
Similar to the results of Vitali et al., none of the nonprion dementias in our series had MRI features suggestive of CJD by any criteria. This criterion has to be validated across other centers for inter-rater and intra-rater reliability along with additional autopsy/histopathological/genetic confirmation to be deemed accurate. MRI positivity in our study was less than what was shown in a recent series of twenty patients wherein all patients had characteristic abnormalities on DWI and/or FLAIR images  as opposed to the mismatch seen in the proposed MRI criteria 2011. However, the superior utility of DWI abnormalities over and above FLAIR hyperintensity has been demonstrated in multiple case series.,,, With concurrent MRI and EEG positivity, only 48.8% patients could be diagnosed in our study. We can only speculate that the diagnostic rates with MRI and EEG could have potentially increased with interval repetition of the investigations. In another recent study, studying the use of MRI for the diagnosis of CJD, the clinical positivity increased to 58.1% and 74.2% using the criteria proposed by the UCSF and the European MRI-CJD Consortium, respectively. In this study, the overt discrepancies between these criteria were thought to be due to the emphasis on clinical examination findings and CSF biomarkers and the highest prevalence (as per European MRI-CJD consortium criteria) was 2.5 times that of the lowest prevalence (as per the WHO criteria using only 14-3-3). Identical diagnostic rates between the UCSF and European MRI-CJD consortium criteria were, however, noted in our study. It is not surprising that with the increase in diagnostic evaluation, as shown with the UCSF and European MRI-CJD consortium data, a center would agree to incur the cost of potentially increasing the false-positive rate. The need to develop in-house facilities for CSF 14-3-3 and other biomarker-based approaches cannot be overemphasized by the fact that three patients of possible CJD did not satisfy the European and UCSF 2007 criteria in the absence of supportive evidence (MRI/EEG) and 11 patients did not meet the proposed UCSF MRI criteria (2011) for diagnosis. This highlights the potential role of CSF and molecular biomarkers as confirmatory tests to reach the highest diagnostic accuracy, although the definitive diagnosis of CJD necessitates the demonstration of the neuropathological triad of neuronal loss, spongiform change, and reactive astrocytosis without an inflammatory reaction. One needs to, however, realize the limitations of CSF 14-3-3 estimation in terms of its high tendency for false negative results.
Our study had certain anticipated limitations. Lack of autopsy with immunohistochemistry for demonstration of protease-resistant prion protein (PrP sc) or genetic data to confirm the diagnosis, was a major limitation of the study. This limited our analysis for assessing the sensitivity of various criteria in the absence of a gold standard. Despite these limitations, our study has its own importance in being one of the largest case series of 'probable' and 'possible' CJD reported from a single center in India with the previous studies reporting not more than ten patients from each center., Special emphasis on the potential advantages as well as limitations of MRI and its applicability in the absence of ante-mortem CSF assays, as well as molecular, or genetic confirmation reflects its utility in centers with resource constraints. We cannot, however, exclude the role of future advances in multimodal neuroimaging in impacting the noninvasive confirmation of CJD.,,
Significant clinical heterogeneity exists in the presentation of CJD making the diagnosis a challenging and exhaustive exercise. This study emphasizes the impact of radiological criteria, with stringent assessment of MRI DWI and FLAIR hyperintensities, in influencing the diagnosis of 'probable' and 'possible' CJD phenotypes. This study also highlights the difficulties encountered in making a diagnosis of 'definite' CJD relying on radiological and EEG biomarkers alone (which is critical in resource-limited institutions) without a histological or molecular confirmation.
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.
[Figure 1], [Figure 2], [Figure 3], [Figure 4]
[Table 1], [Table 2], [Table 3]