Introduction
There are no clear diagnostic guidelines on the role of FDG PET in the diagnostic work-up of cognitive disorders and dementia. Therefore, the European Association of Nuclear Medicine (EANM) and the European Academy of Neurology (EAN) came together in a joint initiative to provide guidance to clinicians on the use of FDG PET in the context of neurodegenerative diseases. The initiative included a set of 21 clinical scenarios, captured as PICO (Population, Intervention, Comparison, Outcome) questions, that were addressed based on literature evidence and expert consensus [
1].
In this article, we focus on the assessment of the quality of studies investigating the clinical utility of FDG PET for identifying patients with Parkinson’s disease (PD) who are at risk of cognitive decline, and the utility of FDG PET in facilitating the differential diagnosis of common forms of parkinsonism, i.e. idiopathic PD, including the prodromal stage, progressive supranuclear palsy (PSP) syndromes, and corticobasal syndrome (CBS). Multiple system atrophy (MSA), another neurodegenerative disorder presenting with parkinsonism was not included in this review. MSA very seldom, if ever, affects cognition and therefore it did not fall within our prespecified condition restricting this review to studies on the role of FDG-PET in “the diagnostic work-up of cognitive disorders and dementia”.
Parkinson’s disease
PD is a common degenerative disorder. It is associated with an increased incidence of cognitive impairment and dementia. The pathology underlying this cognitive decline is variable: while in some patients it is purely Lewy body pathology, in many it is due to mixtures of amyloid, tau and Lewy body pathology [
2]. Any therapeutic intervention to stop cognitive decline is likely to be most effective in the early stages of PD or even during prodromal stages of PD. Consequently, it may become important to identify patients at high risk of cognitive decline before its onset. Older age, scores from nonmotor assessments, reduced dopamine transporter uptake in the caudate, deficit on smell testing, CSF amyloid β (Aβ42) to t-tau ratio, and APOE ε4 status are all known risk factors for cognitive decline in patients with newly diagnosed PD [
3]. Here we report on the role of FDG PET in predicting cognitive decline in PD.
Not all patients presenting with parkinsonism have idiopathic PD. There are alternative pathologies that can present with parkinsonism and cognitive decline. Both PSP and CBS can mimic PD in early stages and are particularly difficult to diagnose in prodromal stages.
Progressive supranuclear palsy
The underlying neuropathology of PSP is a characteristic four-repeat tau neuropathology [
4]. There are a number of clinical phenotypes of PSP which consist of different combinations of motor, gait, language, cognitive and behavioural features [
5]. The two most common phenotypes are PSP Richardson’s syndrome (PSP-RS) and the clinical phenotype that most closely mimics idiopathic PD, PSP-parkinsonism (PSP-P) which often presents with asymmetrical tremor, bradykinesia and rigidity. An initial positive response to levodopa treatment can be misleading. Frequently only later in the disease course, when patients develop additional features of impaired ocular movements with vertical supranuclear gaze palsy, is a retrospective diagnosis of PSP-P made.
Corticobasal syndrome
CBS is an atypical parkinsonian syndrome which consists of dystonia, rigidity, akinesia, myoclonus, tremor and poor response to levodopa. Typically, there is quite marked asymmetry, including limb apraxia and the alien limb phenomenon. Other features include speech and language impairment and cognitive decline. The term CBS describes a clinical phenotype which has been shown to have a heterogeneous underlying pathology. Corticobasal pathology is found only in about 50% of all clinically diagnosed patients. This has led to the distinction between the clinical syndrome (CBS) and the pathological diagnosis (corticobasal degeneration, CBD). The other pathologies found at autopsy include Alzheimer’s disease (AD), PSP and other tauopathies, dementia with Lewy bodies and Creutzfeldt-Jakob disease. Based on this background, three literature searches were performed to assess the quality of evidence supporting the use of FDG PET in facilitating the diagnosis of PD and atypical parkinsonism associated with dementia where the underlying pathology is PSP or CBD.
Methods
Seven panellists, four from the EANM and three from the EAN, were appointed to produce recommendations, taking into consideration the incremental value of FDG PET as an add-on investigation to a comprehensive clinical/neuropsychological assessment, in facilitating the diagnosis and management of patients with parkinsonism. Consensus recommendations were reached through a Delphi procedure which was based on the expertise of the panellists. The panellists were provided with comprehensive data regarding the availability and quality of evidence, which was assessed by an independent methodology team, as described by Boccardi et al. [
6]. Briefly, we searched the literature using harmonized PICO (Population, Intervention, Comparison, Outcome) questions. Thematic keywords were generated by experts, and studies were selected based on eligibility criteria described elsewhere [
6]. Relevant data were extracted from selected studies and assessed for quality of methodology, according to European Federation of Neurological Societies guidance [
7] and for their relevance to FDG PET studies [
6].
PICO questions for this review
In this review, the PICO questions asked whether “performing FDG PET would add diagnostic value (in terms of increased accuracy compared with neuropathological diagnosis, biomarker-based diagnosis or diagnosis at follow-up) to standard clinical/neuropsychological assessment alone”, to:
-
Identify brain dysfunction related to cognitive deterioration in patients with PD and cognitive impairment (PICO question 12)
-
Discriminate PSP from PD (PICO question 13)
-
Identify the underlying pathological process in patients with CBS (PICO question 15).
Eligibility criteria
Only original full papers published in English in international journals with an impact factor were considered. Reviews, management guidelines, abstracts and ‘grey’ literature were excluded. Any sample size was permissible if pathology was the gold standard for diagnosis. Otherwise, the minimum sample size was five for PSP and CBS, and 20 for PD.
Literature search
Electronic searches were performed using a harmonized keywords string based on the specific PICO question, and included a selection of terms chosen for being largely inclusive to identify a broad range of papers. The strings contained a common part for FDG PET and a PICO question-specific part [
6]. The MEDLINE, Embase and Google Scholar databases were searched for studies published up to November 2015. The first screening of all included studies was performed by the panellist responsible for that PICO question (an expert neurologist, psychiatrist or nuclear physician) who could include additional studies based on personal knowledge or forward tracking from the references of papers. The full text of these potentially eligible studies was then independently assessed for eligibility by the methodology team.
Data extraction and quality assessment
To assess the evidence, data on study features, population of interest, index test, gold/reference standard, and critical/proxy outcomes were extracted (for more details see, Boccardi et al. [
6]). Data assessors for this review were D.A. for PICO question 12, and F.G. and S.O. for PICO questions 13 and 15. Critical outcomes were the validated measures of test performance accuracy, sensitivity, specificity, AUC, positive and negative predictive values (PPV, NPV) and likelihood ratios (LR). Additional proxy outcomes for PICO questions 13 and 15 were accuracy of differential diagnosis between typical and atypical parkinsonism. The prediction of years of survival was an additional outcome specific to PICO question 15.
The quality of evidence was consensually assessed by the methodology team based on the study design, gold/reference standard, FDG PET image assessment method (visual or semiquantitative), risk of bias, index test imprecision, applicability, effect size, and effect inconsistency. A final assessment of the relative availability of evidence was formulated, taking into account evidence available from all 21 PICO questions. This ranking was summarized as ‘very poor/lacking’, ‘poor’, ‘fair’ or ‘good’. For further details about data extraction and quality assessment see Boccardi et al. [
6].
Discussion
We assessed the evidence for the clinical utility of FDG PET in diagnosing idiopathic PD and atypical parkinsonism associated with dementia. The evidence supporting the clinical use of FDG PET for identifying PD-related neurodegeneration associated with impaired cognition and for discriminating PSP from idiopathic PD was lacking. Conversely, the evidence supporting the clinical use of FDG PET for identifying the underlying neuropathology in patients with CBS was fair. Despite the low quality of evidence available from published studies, the Delphi panel voted for recommending the clinical use of FDG PET in the differential diagnosis of conditions characterized by parkinsonism and dementia (PICO question 12).
The reasons given by panellists during the Delphi panel process focused on both the clinical utility of the NPV of FDG PET, and also on the PPV of typical patterns of hypometabolism. In particular for the identification of impending PD-related cognitive decline, the panellists recommended the clinical use of FDG PET because patients with PDD or PD-MCI have a typical pattern of hypometabolism mainly affecting the posterior cortical areas. The prognostic value of FDG PET was also considered to be clinically relevant in the identification of patients who may benefit from early cholinesterase inhibitor treatment or other future symptomatic treatments. A recently published study [
67] showed that FDG PET with statistical parametric mapping detected patterns of hypometabolism that predicted the risk of a patient with PD having progressed to dementia by 4 years with 85% sensitivity and 88% specificity. However, this study was not available at the time of the literature search and therefore was not considered in the Delphi process. Atypical hypometabolism of mainly posterior cortical areas on FDG PET could therefore be added to the list of other helpful investigations which include reduced dopamine transporter uptake in the caudate, CSF amyloid β (Aβ1-42) to t-tau ratio and APOE ε4 status [
3], and is a risk factor for cognitive deterioration in idiopathic PD.
For PICO question 13, panellists based their decision to support the clinical use of FDG PET for discriminating PSP from idiopathic PD on the presence of a typical metabolic pattern for PSP, which is not present in PD. PSP is characterized by hypometabolism in the medial frontal and anterior cingulate cortices, and in the striatum and midbrain. FDG PET may therefore be useful in early stages of the disease, when the clinical diagnosis is less certain. Perfusion SPECT, albeit less precise because of poorer spatial resolution, also displays a consistent pattern [
68,
69]. The described abnormalities in PSP can be difficult to detect in very early stages by visual analysis alone and a semiautomated assessment, comparing the pattern in the patient with the pattern in age-matched controls is recommended in addition to visual reading, consistent with recommendations of the present EANM-EAN initiative [
70]. While a PSP-related pattern has been repeatedly demonstrated [
33,
62,
71], data are still incomplete if different PSP phenotypes are considered, such as PSP-P or pure akinesia with gait freezing. These may be characterized by less severe or incomplete patterns, compared with typical and full-blown PSP. The panellists’ decision was consistent with both the EANM procedural guidelines [
72] and the more recent diagnostic criteria of the Movement Disorder Society [
5,
73], which support the use of FDG PET for discriminating PD from atypical parkinsonian syndromes.
Regarding CBS (PICO question 15), an asymmetrical cortical hypometabolism is known to affect the hemisphere contralateral to the side with akinetic-rigid parkinsonism and apraxia. This hypometabolism is typically found in the motor and premotor cortices, but may also involve the prefrontal or posterior parietal and lateral temporal cortex, and the cingulate gyrus. The heterogeneity of metabolic patterns found in CBS when no autopsy diagnosis is available is most likely due to the variety of different pathologies that can present as CBS. These include CBD, PSP, AD and frontotemporal dementia, and a mix of these conditions. The future challenge is to differentiate the metabolic patterns associated with different underlying pathologies and will require either neuropathological confirmation of the diagnosis by autopsy or the use of additional imaging methods, for example tau-PET and amyloid-PET as the gold standard.
Considering the large number of studies published, the fact that the majority had severe limitations and did not allow the generation of evidence that could confidently support the clinical use of FDG PET in the three PICO scenarios was striking. It is important that future studies seek to recruit sufficient numbers of patients and have a robust clinical diagnosis and follow-up and ideally include additional imaging or other biomarkers to strengthen diagnostic certainty. Studies should preferably include autopsy diagnosis as the gold standard. Other important considerations are appropriate clinical comparisons, innovative statistical analysis and the use automatic semiquantitative assessment in addition to visual assessment, to help the nuclear medicine physician interpret findings and to increase specificity and/or sensitivity. In the case of PD, analysing FDG PET using a spatial covariance method could identify characteristic patterns of metabolism in PD with and without cognitive decline. This could be helpful in future research not only to aid more accurate diagnosis but also to evaluate the effect of new therapeutic interventions [
74].
Acknowledgments
The procedure for assessing scientific evidence and defining consensual recommendations was funded by the European Association of Nuclear Medicine (EANM) and by the European Academy of Neurology (EAN). We thank the Guidelines Working Group of EAN, particularly Simona Arcuti and Maurizio Leone, for methodological advice.
Compliance with ethical standards
Conflicts of interest
Zuzana Walker received grants and tracers, personal fees for consultancy and a speaker’s fee from G.E. Healthcare.
Federica Gandolfo declares that she has no conflicts of interest.
Stefania Orini declares that she has no conflicts of interest.
Valentina Garibotto declares that she has no conflicts of interest.
Federica Agosta is Section Editor of NeuroImage: Clinical; has received speaker fees from Biogen Idec, Novartis, and Excellence in Medical Education; and receives or has received research support from the Italian Ministry of Health, AriSLA (Fondazione Italiana di Ricerca per la SLA), and the European Research Council. She received personal fees from Elsevier INC.
Javier Arbizu received grants from Eli-Lilly & Co, Piramal and GE Healthcare.
Femke Bouwman declares that he has no conflicts of interest.
Alexander Drzezga received grants and nonfinancial support from Eli-Lilly & Co, Siemens and GE Healthcare; he also received nonfinancial support from Piramal.
Peter Nestor declares that he has no conflicts of interest.
Marina Boccardi received funds from the European Association of Nuclear Medicine (EANM) to perform the evidence assessment and the global coordination of the present project. She also received research grants from Piramal and has served as a paid member of advisory boards for Eli Lilly.
Daniele Altomare received a grant allocated by the European Academy of Neurology (EAN) for data extraction and evidence assessment for the present project.
Cristina Festari declares that she has no conflicts of interest.
Flavio Nobili received personal fees and nonfinancial support from GE Healthcare, nonfinancial support from Eli-Lilly and grants from Chiesi Farmaceutici.
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.