A Cross-Validation of FDG- and Amyloid-PET Biomarkers in Mild Cognitive Impairment for the Risk Prediction to Dementia due to Alzheimer’s Disease in a Clinical Setting

Abstract

Assessments of brain glucose metabolism (¹⁸F-FDG-PET) and cerebral amyloid burden (¹¹C-PiB-PET) in mild cognitive impairment (MCI) have shown highly variable performances when adopted to predict progression to dementia due to Alzheimer’s disease (ADD). This study investigates, in a clinical setting, the separate and combined values of ¹⁸F-FDG-PET and ¹¹C-PiB-PET in ADD conversion prediction with optimized data analysis procedures. Respectively, we investigate the accuracy of an optimized SPM analysis for ¹⁸F-FDG-PET and of standardized uptake value ratio semiquantification for ¹¹C-PiB-PET in predicting ADD conversion in 30 MCI subjects (age 63.57±7.78 years). Fourteen subjects converted to ADD during the follow-up (median 26.5 months, inter-quartile range 30 months). Receiver operating characteristic analyses showed an area under the curve (AUC) of 0.89 and of 0.81 for, respectively, ¹⁸F-FDG-PET and ¹¹C-PiB-PET. ¹⁸F-FDG-PET, compared to ¹¹C-PiB-PET, showed higher specificity (1.00 versus 0.62, respectively), but lower sensitivity (0.79 versus 1.00). Combining the biomarkers improved classification accuracy (AUC = 0.96). During the follow-up time, all the MCI subjects positive for both PET biomarkers converted to ADD, whereas all the subjects negative for both remained stable. The difference in survival distributions was confirmed by a log-rank test (p = 0.002). These results indicate a very high accuracy in predicting MCI to ADD conversion of both ¹⁸F-FDG-PET and ¹¹C-PiB-PET imaging, the former showing optimal performance based on the SPM optimized parametric assessment. Measures of brain glucose metabolism and amyloid load represent extremely powerful diagnostic and prognostic biomarkers with complementary roles in prodromal dementia phase, particularly when tailored to individual cases in clinical settings.

Keywords

Alzheimer’s disease conversion prediction dementia early diagnosis mild cognitive impairment prognosis

INTRODUCTION

The mild cognitive impairment (MCI) construct is nowadays the most clinically valid entity to define a stage of possible transition from moderate cognitive impairment in aging to dementia. However, the trajectories of cognitive decline and the conversion to different forms of dementia call for the identification of optimal biomarkers for outcome prediction [1]. The risk prediction of conversion to dementia due to Alzheimer’s disease (ADD) has crucial implications for the best fitting in clinical trials of subjects with correct AD diagnosis, particularly in prodromal and preclinical phases [2, 3]. Decreases of cerebral glucose metabolism in temporo-parietal and precuneus regions as measured by ¹⁸F-FDG-PET, and the amyloid burden, as detected with amyloid-PET, are adopted in the National Institute of Aging-Alzheimer Association (NIA-AA) criteria for both dementia and MCI due to AD pathology [4, 5]. According to these criteria, ¹⁸F-FDG-PET hypometabolism and amyloid-PET positivity, reflecting respectively downstream neuronal degeneration and pathological amyloid deposition, are both considered in vivo supporting biomarkers for AD pathology and are required to establish diagnosis [4, 5]. In contrast, the International Working Group criteria (IWG-2) have proposed a simplified algorithm in which only cerebrospinal fluid (CSF) measurements of amyloid and tau protein or amyloid-PET status are considered as supportive diagnostic biomarkers of AD pathology [6], while ¹⁸F-FDG-PET is considered only for disease staging and monitoring of the progression [6]. A recent study on the prognostic value of the NIA-AA [5], IWG-2 and IWG-1 criteria [6, 7], however, showed that classification according to the NIA-AA criteria, in which PET pathological and neuronal injury biomarkers are both considered for AD diagnosis, reached the highest prediction accuracy for MCI to ADD conversion in a clinicalsetting [8].

As for prediction of conversion to ADD in MCI subjects, both ¹⁸F-FDG-PET and amyloid-PET techniques have shown high variability in their performances, as recently reported in two Cochrane reviews [9, 10]. As for the former [9], the ¹⁸F-FDG-PET meta-analysis showed high variability in sensitivity and specificity, thus advising against ¹⁸F-FDG-PET routine utilization in MCI subjects [9]. A critical stance by the European Association of Nuclear Medicine (EANM) followed [11]; highlighting the limitations of the Cochrane review and claiming that the observed variability was mostly due to confounding biases, among them, crucially, the methodology to define ¹⁸F-FDG-PET scan positivity [11]. Similarly, the Cochrane review on ¹¹C-PiB-PET for risk assessment in MCI and differential diagnosis of dementias [10] concluded for a considerable heterogeneity in the conduct and interpretation of the results, particularly due to the lack of defined thresholds for test positivity and standardization of analysis methods, thus not recommending amyloid PET in routine clinical practice [10]. Other sources of variability for ¹¹C-PiB-PET results relate crucially to the different follow-up durations employed by different studies, as well as to sampling bias related to varying operational MCI definitions, accordingly to a recent review [12].

Thus, the adoption of validated and standardized methods for ¹⁸F-FDG-PET and amyloid-PET in research and clinical settings is fundamental to achieve high accuracy, as shown in two large recent meta-analyses [13, 14]. This is especially true since the role for ¹⁸F-FDG-PET in the early and differential diagnosis of dementias and for amyloid PET in excluding the presence of underlying AD pathology is confirmed by several reports [15 –21].

In a sample of MCI subjects evaluated in a clinical setting, we investigated the separate and combined performances of ¹⁸F-FDG-PET and ¹¹C-PiB-PET imaging, respectively, analyzed with an optimized SPM procedure [22, 23] and semi-quantified standardized uptake value ratio (SUVR) methods [24], in predicting or excluding ADD conversion at the individual level. A follow-up period confirmed the ADD conversion or the stable MCI condition.

METHODS

Participants

We included 30 MCI subjects, classified according to diagnostic criteria [1], all of them retrospectively included from the database of the Karolinska Institutet, Stockholm, Sweden. Inclusion criteria for the study were the availability of both ¹¹C-PiB-PET and ¹⁸F-FDG-PET imaging and a clinical follow-up to evaluate the conversion to ADD. MCI clinical subtypes were assessed following standardized criteria [1], thus splitting the whole sample in amnestic/non-amnestic, single- or multiple-domain MCI (aMCI SD, aMCI MD, naMCI SD, naMCI MD, see Table 1). Progression from MCI to ADD was established according to standardized clinical diagnostic criteria [5]. At the clinical follow-up, 14 MCI converted to ADD (median time to conversion 21.5, inter-quartile range IQR 34.8 months) and 16 MCI remained stable (median 31.5 months, IQR 25.8 months). Demographic and clinical summary for the whole cohort and split by conversion status is available in Table 1.

Table 1

Demographic and clinical summary of the MCI cohort

Index	Whole Sample	Converter MCI cases	Non-converter MCI cases	p-value
	(n = 30)	(n = 14)	(n = 16)
Gender (n Females)	20	10	10	0.709
APOE ɛ4 carriers pos/neg	20/10	12/2	8/8	0.058
Age in years (mean±sd)	63.57±7.78	60.93±7.39	65.87±7.59	0.052
Education in years (mean±sd)	13.03±3.18	13.08±3.52	13.00±2.99	0.714
MMSE at baseline (mean±sd)	27.63±1.56	27.14±1.51	28.06±1.53	0.110
Follow-up in months (median/IQR)	26.5/30	21.5/34.8	31.5/25.8	0.313
Clinical Presentation (N and %)
aMCI SD	18 (60%)	7 (50%)	11 (68.8%)
aMCI MD	6 (20%)	3 (21.4%)	3 (18.7%)	0.429
naMCI SD	4 (13.3%)	2 (14.3%)	2 (12.5%)
naMCI MD	2 (6.7%)	2 (14.3%)	0 (0%)

Table showing demographic and clinical data of the studied MCI cohort, split by conversion status. p-values represent the significance of pairwise converter versus non-converter MCI group comparisons. Fisher Exact Test (two-tailed) was adopted for the bimodal discrete variables, otherwise a two-tailed χ ² test was performed. APOE, Apolipoprotein E; MMSE, Mini-Mental State Examination; IQR, interquartile range; aMCI SD, amnestic MCI single-domain; aMCI MD, amnestic MCI multi-domain; naMCI SD, non-amnestic MCI single-domain; naMCI MD, non-amnestic MCI multi-domain.

All subjects, or their informants/caregivers, gave written informed consent to the experimental procedure that had been approved by the regional Human Ethics Committee of Stockholm as well as the by the radiation safety committee of Uppsala University Hospital.

Cognitive assessment

Global cognitive functioning at baseline was assessed using Mini-Mental Status Examination (MMSE). Multiple cognitive domains were evaluated adopting an extended standardized neuropsychological battery. The neuropsychological tests included the Swedish versions of the Wechsler Adult Intelligence Scale (WAIS), including the Full Scale Intelligence Quotient (FSIQ), Information, Similarities and Block Design subscales. Additionally, verbal short-term memory (Digit Span, Rey Auditory Verbal Learning Test – RAVLT immediate recall), short-term memory, attention and processing speed (Digit Symbol), visuo-constructive abilities (Rey Figure copy), long-term episodic memory (RAVLT delayed recall), long-term visual spatial memory (Rey Figure recall), and visuo-motor, executive and attention functioning (Trail Making Test – A/B) tests. In order to analyze neuropsychological measures, raw scores were transformed into z-scores according to the national normative data [25].

PET Studies

¹⁸F-FDG-PET acquisition

¹⁸F-FDG-PET scans were acquired either with a Siemens ECAT EXACT HR+ or a GEMS Discovery ST PET/CT scanner with 3D mode yielding a 155–157 mm axial field of view. The injected dose was 214±45 MBq, and was acquired through a 3D dynamic scan of either 60 (5×60, 5×180, 6×300, 1×600 s) or 45 min (1×60, 1×1140, 5×300 s / 4×30, 9×60, 3×180, 5×300 s). Image reconstruction protocol followed a 3D Filtered Back Projection (FBP). Attenuation correction was performed with rotating 68Ge rod sources for the ECAT EXACT HR+ scanner whereas it was CT-based for the GEMS Discovery-ST. All reconstructed frames were re-aligned to correct for between frame patient motions (VOIager).

¹⁸F-FDG-PET data analysis

Image analysis was carried out with SPM5 software (Wellcome Department of Imaging Neuroscience, London, UK; www.fil.ion.ucl.ac.uk/spm) on MATLAB 8 (MathWorks Inc, Sherborn, Mass). The late 15-min summation image of each ¹⁸F-FDG-PET scan was analyzed according to an optimized ¹⁸F-FDG-PET SPM procedure previously developed and validated in early and differential dementia diagnosis [18 , 26–29] and in prodromal MCI outcome prediction [17]. Briefly, scans were ‘spatially normalized’ in accordance to a reference “dementia specific” ¹⁸F-FDG-PET template [22] and then smoothed 8 mm. These single-subject images were compared to a normality database tool by means of a two-sample t-test implemented in SPM5 to assess brain hypometabolism (detailed description available in [22, 23]. The statistical threshold was set at p = 0.05, family wise error (FWE) corrected for multiple comparisons. Only clusters containing more than 100 voxels were deemed to be significant. This method has been shown to be valid, accurate and reliable in detecting AD metabolic signature independently from the PET scanner used for image acquisition [30].

¹¹C-PiB-PET acquisition

Amyloid-PET scans were performed using ¹¹C-PiB-PET and either a Siemens ECAT EXACT HR+ scanners or a GEMS Discovery ST PET/CT scanner with 3D mode yielding a 155–157 mm axial field of view. The mean injected dose was 247±72 MBq. The ¹¹C-PIB-PET scans consisted of 24 frames (4×30; 9×60; 3×180; 8×300 s) with a total duration of 60 min with a FBP protocol adopted for image reconstruction. Attenuation correction was performed with rotating 68Ge rod sources for the ECAT EXACT HR+ scanner whereas it was CT-based for the GEMS Discovery-ST. All reconstructed frames were re-aligned to correct for between frame patient motions (VOIager).

¹¹C-PiB-PET data analysis

¹¹C-PiB-PET late 40–60 min summation images were created and nonlinearly spatially normalized to a population-based PIB template in Montreal Neurological Institute (MNI) space [24] with the use of SPM8 software. A cerebellar grey matter region [24] was used as reference for each 40–60 min PIB image to create SUVR (SUVR_CERGM) parametric images.

Neuroimaging biomarkers positivity

¹⁸F-FDG-PET

The single-subject ¹⁸F-FDG-PET SPM hypometabolism maps were evaluated by four experts in neuroimaging, blind to each other and to clinical information and amyloid status. The expert raters showed a high agreement (Cohen’s k: 0.89) in the evaluation of SPM maps. Images were defined as AD-positive pattern (FDG+) when ¹⁸F-FDG-PET SPM hypometabolism maps represented either the typical (temporoparietal and/or posterior cingulate hypometabolism) [18 , 31] or an AD atypical pattern as follows: 1) predominant left posterior perisylvian and parietal hypometabolism, suggestive for the logopenic variant of primary progressive aphasia (lvPPA) [32]; 2) predominant posterior parietal and occipital brain hypometabolism for posterior cortical atrophy (PCA) [26, 33]; and 3) frontal and temporo-parietal hypometabolism with posterior cingulate involvement for the frontal variant AD [34]. Otherwise, the ¹⁸F-FDG-PET SPM hypometabolism maps were defined as AD-negative pattern (FDG–) when they did not show differences in the comparison with normal control database.

¹¹C-PiB-PET

Composite amyloid SUVR scores resulting from the average of uptake values of medial orbitofrontal, superior parietal, middle temporal, precuneus, anterior and posterior cingulate regions were estimated. We adopted template-based regions of interest (ROIs) created through the Wake Forest University PickAtlas toolbox for SPM [35], using anatomical definitions from the automated anatomical labeling (AAL) atlas [36]. These composite SUVR scores were then compared to a validated cut-off threshold, i.e., 1.41 [24]. All the subjects were classified as amyloid positive (Amy+) when their composite cortical SUVR score was above the normality cut-off threshold, and as amyloid negative (Amy–), when it was below the normality cut-off threshold.

Biomarkers analysis for progression prediction

We first evaluated the separate ability of both PET biomarkers in predicting or excluding conversion to dementia, as established at the clinical follow-up diagnosis. We then estimated sensitivity, specificity, accuracy for each PET biomarker, together with a receiver operating characteristic (ROC) analysis with area under the curve (AUC) estimation, which was performed setting clinical status at last follow-up as the primary outcome.

Second, to evaluate the combined diagnostic power of both PET biomarkers, we ran a forward conditional logistic regression, setting ¹⁸F-FDG-PET and ¹¹C-PiB-PET results (FDG+/–and Amy+/–) as predictors and dementia conversion status as outcome. This procedure allows the creation of a third variable in which each subject has a predicted probability value according to the combination of ¹⁸F-FDG-PET and ¹¹C-PiB-PET ratings. We then used this third variable to test the classification ability of the biomarkers combination in predicting conversion to ADD. Considering the group as split by biomarker combination status (e.g., FDG+/Amy+, FDG–/Amy–), we estimated a Kaplan-Meier survival distribution considering ADD conversion as event as a function of the follow-up time. All the statistical analyses were performed with SPSS software (v21.0 http://www.ibm.com/analytics/us/en/technology/spss/).

RESULTS

Cognitive assessment

Converter MCI when compared to the non-converters showed more impairment at baseline in the long-term visual spatial memory task (Rey Figure recall) (Mann-Whitney non-parametric U test, p = 0.011) and in the WAIS Similarities subscale (Mann-Whitney non-parametric U test, p = 0.011). No significant differences were found in the other neuropsychological test scores at baseline between the two MCI subgroups. A summary of neuropsychological data is available in Table 2.

Table 2

Neuropsychological summary of the MCI cohort

Test (mean±sd)	Whole Sample	Converter MCI cases	Non-converter MCI cases	p-value
	(n = 30)	(n = 14)	(n = 16)
MMSE	27.63±1.56	27.14±1.51	28.06±1.53	0.110
WAIS - FSIQ	–0.85±1.27	–1.42±1.32	–0.42±1.09	0.054
WAIS - Information	–0.36±0.96	–0.63±0.91	–0.14±0.98	0.183
WAIS - Similarities	–0.07±1.22	–0.75±1.33	0.44±0.87	0.011*
WAIS - Block Design	–1.04±1.21	–1.43±1.45	–0.75±0.94	0.174
Rey Figure Copy	–0.99±2.25	–1.77±3.15	–0.35±0.76	0.398
Digit Span	–0.23±1.04	–0.01±0.98	–0.43±1.07	0.400
Corsi Span	–1.14±1.27	–1.15±0.96	–1.13±1.57	0.684
RAVLT immediate	–0.95±1.02	–1.23±1.09	–0.73±0.94	0.324
RAVLT delayed	–1.17±1.12	–1.37±0.88	–1.02±1.28	0.450
Rey Figure Recall	–1.16±0.96	–1.71±0.98	–0.75±0.74	0.011*
Digit Symbol	–0.60±1.29	–1.03±1.51	–0.28±1.03	0.259
Trail Making Test –A	–0.95±1.66	–1.52±2.00	–0.50±1.20	0.167
Trail Making Test –B	–1.25±2.32	–1.89±3.11	–0.73±1.32	0.427

Table showing neuropsychological summary of the studied MCI cohort, split by conversion status. Values are shown as z-scores, with reported mean±sd. p-values represent the significance of pairwise converter versus non-converter MCI group comparisons. * Indicates a significant (p < 0.05) difference (Mann-Whitney U-test for independent samples), MMSE, Mini-Mental State Examination; WAIS, Wechsler Adult Intelligence Scale; FSIQ, Full Scale Intelligence Quotient; RAVLT, Rey Auditory Verbal Learning Test. See Methods for details.

Biomarkers positivity

Within the 14 converter MCI cases, 11 were FDG+/Amy+, showing both the typical pattern (n = 4) and atypical AD brain glucose hypometabolism patterns (i.e., 2 lvPPA, 2 PCA, and 1 frontal variant of AD patterns). In addition, there were two subjects with a typical AD hypometabolism pattern with mixed evidence for cerebrovascular disease. Finally, the other three MCI converter cases were FDG–/Amy+. All the 16 non-converter MCI cases were FDG-negative. As regards ¹¹C-PiB-PET, 10/16 were Amy–and 6/16 were Amy+. The results are summarized on Table 3. Figure 1 depicts representative ¹⁸F-FDG-PET for typical AD, lvPPA, and PCA patterns with the respective ¹¹C-PiB SUVRs maps.

Table 3

PET biomarker combinations and conversion to AD dementia

	Whole Sample	Converter MCI cases	Non-converter MCI cases
	(n = 30)	(n = 14)	(n = 16)
FDG+/Amy+ (% N)	11 (37%)	11 (79%)	0 (0%)
FDG–/Amy+ (% N)	9 (30%)	3 (21%)	6 (37%)
FDG–/Amy–(% N)	10 (33%)	0 (0%)	10 (63%)

Table showing combinations of biomarkers positivity and AD conversion in our MCI cohort. Group sizes are also shown. FDG+, FDG pattern suggestive for AD; FDG–, FDG pattern negative for AD; Amy+, amyloid composite ¹¹C-PiB-PET SUVR above positivity cut-off; Amy–, amyloid composite ¹¹C-PiB-PET SUVR below positivity cut-off.

Fig.1

Optimized ¹⁸F-FDG-PET SPM maps and ¹¹C-PiB-PET SUVRs in representative MCI cases. Figure showing optimized ¹⁸F-FDG-PET SPM maps and ¹¹C-PiB-PET SUVR_CERGM images of representative MCI subjects split by clinical outcome and biomarkers positivity combinations. Diverging ¹⁸F-FDG-PET hypometabolism SPM patterns representative of the AD variants are associated with comparable ¹¹C-PiB-PET amyloid retention distributions. Optimized ¹⁸F-FDG-PET SPM and ¹¹C-PiB-PET SUVR_CERGM images are overlaid on a T1 standard anatomical template with MRIcron software (http://www.mccauslandcenter.sc.edu/crnl/tools).

Biomarkers progression prediction

Both modalities showed valuable prediction ability, with ¹⁸F-FDG-PET positivity performing as the best individual predictor for AD conversion, showing an overall higher accuracy than ¹¹C-PiB-PET (0.90 versus 0.80, respectively). ¹⁸F-FDG-PET showed higher specificity than ¹¹C-PiB-PET (1.00 versus 0.62), but lower sensitivity (0.79 versus 1.00). AUCs were 0.89 for ¹⁸F-FDG-PET and 0.81 for ¹¹C-PiB-PET. Adding genetic risk, i.e., APOE status, to either the PET-based measurements did not improve the prediction, or exclusion, of conversion to ADD.

Noteworthy, ¹⁸F-FDG-PET and ¹¹C-PiB-PET combined showed the highest AUC (0.96) (see Fig. 2A). None of the FDG–/Amy–subjects converted to AD, leading to 1.0 probability of non-conversion at about 10 years from MCI diagnosis. On the other hand, all the FDG+/Amy+ converted within about 8 years of follow-up. Lastly, three FDG-/Amy+ subjects converted to ADD either very early after initial MCI diagnosis (n = 2, converting after 6 and 7 months of follow-up) or about 7.5 years later (n = 1, converting after 90 months of follow-up). Kaplan-Meier curves showed a clear difference between the survival probability of the FDG+/Amy+, FDG–/Amy+, and FDG–/Amy–groups (see Fig. 2B). The right censoring pattern differed across the biomarker groups, in particular with lack of censored events for the FDG+/Amy+ group and complete censoring for the FDG–/Amy–. A log-rank (Mantel-Cox) test confirmed the statistically significant difference between the survival distributions of the three groups (χ ²: 12.051, p = 0.002). The log-rank test was preferred since it focuses on the longer survival times and is more resistant to differences in the right censoring patterns, which are in contrast with the proportional hazard assumption of the Kaplan-Meier test.

Fig.2

ROC and Kaplan-Meier curves for ADD conversion prediction. Figure showing ROC and Kaplan-Meier Survival Curves for ADD conversion prediction. A) ROC curves for the individual ¹⁸F-FDG-PET (violet) and ¹¹C-PiB-PET (blue) biomarkers, together with their combination (orange). The dashed black line represents a diagonal reference. AUCs were: 0.89 for ¹⁸F-FDG-PET, 0.81 for ¹¹C-PiB-PET and 0.96 for their combination. B) Image showing Kaplan-Meier survival plots split by subgroups derived from PET biomarkers positivity combinations, respectively: FDG-/Amy–(green), FDG-/Amy+ (red), and FDG+/Amy+ (blue). Dashes represent right censored cases. See text for details.

DISCUSSION

In this study we investigated, in single MCI individuals, the separate and combined performances of ¹⁸F-FDG-PET and ¹¹C-PiB-PET measures, using validated semi-quantified methods, as biomarkers of neurodegeneration and pathology status able to exclude or predict the risk of conversion to ADD. From the clinical and neuropsychologicalstandpoint, at baseline, the converter MCI performed significantly worse than the stable subjects in long-term visual spatial memory (Rey Figure recall) and productive/receptive language (WAIS Similarities subscale). These findings are coherent with the known AD-related neuropsychological signature [5].

The prediction analyses showed partly different performances for topographical (¹⁸F-FDG-PET) and pathological (¹¹C-PiB-PET) biomarkers, with higher specificity for ¹⁸F-FDG-PET SPM maps and higher sensitivity for ¹¹C-PiB-PET SUVRs. In details, the conversion rate of ¹⁸F-FDG-PET-positive subjects was 100%, whereas the conversion rate of PET amyloid-positive subjects was 70%, consistently with previous literature [24, 37]. All the subjects positive for both ¹⁸F-FDG-PET and ¹¹C-PiB-PET (n = 11) were diagnosed with ADD during the follow-up period. The combination of these PET biomarkers showed an increase in overall accuracy for ADD conversion prediction in MCI subjects (Fig. 2A).

In this series, ¹⁸F-FDG-PET SPM maps were able to detect not only the typical AD hypometabolism signature, but also the hypometabolism patterns suggestive of atypical clinical presentation of AD pathology (i.e., lvPPA, PCA, and frontal variant) [34] (see Fig. 1). The optimized ¹⁸F-FDG-PET SPM method here confirmed previous evidence on its effective role in differentiation of atypical clinical phenotypes associated with AD pathology [26, 28]. There is a consistent body of evidence regarding the potential of ¹⁸F-FDG-PET in detecting specific hypometabolism patterns across different dementia conditions, e.g., ADD, dementia with Lewy bodies, and frontotemporal dementia [17 , 38]. The previous literature and present results strongly advocate for a crucial role of topographical downstream neurodegeneration markers, such as ¹⁸F-FDG-PET, particularly in MCI conversion prediction and also in early dementia classification, with critical implications in the use of voxel-wise semiquantitative methods. Additionally,few ¹⁸F-FDG-PET SPM maps were rated as representative of mixed AD plus cerebrovascular disease (see Results). Neuroimaging-based biomarkers have increasingly been recognized as of value for clinical practice in the diagnosis of underlying vascular disease [39, 40]. Structural magnetic resonance imaging (MRI) techniques are the standard choice when cerebrovascular disease is suspected, revealing specific imaging abnormalities associated with vascular events [41]. ¹⁸F-FDG-PET studies in cerebrovascular diseases have revealed typical patterns of focal scattered and patchy hypometabolism in cortical and subcortical regions [39]. Of note, ¹⁸F-FDG-PET measurement could add to MRI the evidence of dysfunctional areas, due to its sensitivity in detecting cortical deafferentation and synaptic dysfunction in normal-appearing tissue [40].

Crucially, all the subjects negative for both ¹⁸F-FDG-PET and ¹¹C-PiB-PET remained stable over the follow-up time. This finding supports the fundamental value of both biomarkers in excluding neurodegenerative process due to underlying AD pathology in the MCI stage. This high negative ADD risk prediction was driven by the high value of the ¹⁸F-FDG-PET negativity to rule out AD-associated neurodegeneration in MCI cases, as previously found and discussed [17]. The high specificity of ¹⁸F-FDG-PET in ADD, dementia with Lewy bodies, and frontotemporal dementia implies that a negative, or normal scan in the presence of the suspicion of dementia makes a diagnosis of a neurodegenerative disease very unlikely. In the context of early diagnosis, the exclusionary role of ¹⁸F-FDG-PET is especially important in younger subjects with a suspicion of neurodegenerative disease [20]. With regards to amyloid-PET, its potential in ruling out AD pathology is well known [42].

Six out of 16 non-converter MCI showed an Amy+ pattern with a negative ¹⁸F-FDG-PET scan, at a median follow-up time of 25 months. We suggest that these FDG–/Amy+ non-converter MCI cases could either be in need of longer follow-ups or, when remaining stable over long time, could be considered as incidental amyloid cases [43]. In addition, as previously stated, the presence of high PET amyloid burden in MCI subjects does not provide a sufficient confidence of progression to ADD [43, 44]. A known limit of amyloid-PET imaging regards the prevalence of subjects, increasing with age, who do not develop dementia or do not present cognitive alterations in spite of a significant amyloid burden [43]. For these subjects, the topographical ¹⁸F-FDG-PET downstream biomarker fully shows its potential in ruling out ongoing neurodegenerative process in vivo. ¹⁸F-FDG-PET imaging has been indeed shown to have a valuable exclusionary role, thus really bearing a crucial added prognostic value in MCI [13 , 20]. Three out of 14 converter MCI cases were FDG negative but Amyloid positive, presenting, respectively, 6, 7, and 90 months of follow-up. This evidence supports a possible temporal dissociation between the two PET biomarkers, with an earlier positivity of amyloid-PET imaging in some cases, which supports the need of combined adoption of pathological and neurodegenerative biomarkers, when available.

Only few studies evaluated the impact of ¹⁸F-FDG-PET and ¹¹C-PiB-PET in the same cohort of MCI subjects, although with inconsistent results, especially with regards to ¹⁸F-FDG-PET imaging [46 –49]. Overall, these studies showed lower performances of ¹⁸F-FDG-PET in the MCI to ADD conversion prediction, which is likely to be due to variability in follow-up durations and, more importantly, to the heterogeneity in the analysis methodology [46 –49]. The ¹⁸F-FDG-PET quantification methods varied from ¹⁸F-FDG-PET SUVR composite scores [46, 48], to the use of fully automated analyses [47, 49]. The higher sensitivity and specificity of ¹⁸F-FDG-PET found in the present study is likely related to 1) follow-up lengths, 2) the use of an optimized and validated SPM method for the analysis of ¹⁸F-FDG-PET data with expert image interpretation [23], and 3) the study design. As for the latter, in our retrospective study the MCI subjects within the internal database of a clinical research center are representative of a highly-selected group, not reflecting the standard consecutive population observed in a memory clinic. Even if the present results are specific to this experimental clinical scenario, the high accuracy of the optimized ¹⁸F-FDG-PET SPM method in this MCI group was comparable to the accuracy level obtained in more heterogeneous cohorts of MCI subjects, in predicting, or excluding, conversion to different dementia syndromes [17 , 23].

Conclusion

To date, there are no absolute timing algorithms for diagnosis of ADD providing clear recommendation for the use of ¹⁸F-FDG-PET and amyloid-PET imaging in a clinical setting [9, 10]. ¹⁸F-FDG-PET as a biomarker of AD- as well as non-AD neurodegeneration has shown a very high specificity, which is of utmost relevance for differential diagnosis in the early phase of dementia and for the exclusion of underlying neurodegenerative processes. On the other hand, amyloid-PET represents a unique biomarker with exceptional sensitivity for in vivo investigation of underlying cerebral AD pathology.

There is some added value for the combined use of these PET biomarkers, especially when adopted with validated quantification approaches as shown in the present study, given that the combination of these techniques improved the global accuracy of ADD conversion prediction. We suggest, however, that the recommendation for adoption of both PET techniques for MCI prognosis should be tailored on the individual cases, taking into account clinical, neuropsychological, and the availability of other biomarker data.

Finally, this work contributes to the need for standardization of core biomarkers collection and measurements and the urgency to develop evidence-based guidelines for high-level accuracy biomarkers, useful for early dementia diagnosis, and in particular for the inclusion or exclusion in prevention clinical trials [50].

Footnotes

ACKNOWLEDGMENTS

L.I. holds a ministerial fellowship granted by the Italian Ministry of Education, University and Research (Graduate program). The research received funding from the Italian Ministry of Health (Ricerca Finalizzata Progetto Reti Nazionali AD NET-2011- 02346784), from the EU FP7 INMIND Project (FP7-HEALTH-2011-two-stage “Imaging of Neuroinflammation in Neurodegenerative Diseases”, grant agreement no. 278850) and from the Joint Program on Neurodegenerative Disorders (EU-JPND) project Biomarkers for Alzheimer’s Disease and Parkinson’s Diagnostics (BIOMARKAPD).

K.C and A.N. were financially supported by the Swedish Research Council (project 05817), Swedish Brain Power, the Regional Agreement on Medical Training and Clinical Research (ALF) for Stockholm County Council, the Strategic Research Programme in Neuroscience at Karolinska Institutet, the Foundation for Old Servants, Gun and Bertil Stohne’s Foundation, KI Foundations, the Swedish Brain Foundation, the Swedish Alzheimer’s Foundation (Alzheimerfonden), Demensfonden, the Wenner-Gren Foundation, the European Union’s Seventh Framework Programme (FP7/2007-2013) under grant agreement n° HEALTH-F2-2011-278850 (INMiND), and the Swedish Foundation for Strategic Research (SSF).

KC and AN wish to thank Mr. Johan Lilja for support related to the imaging software VOIager. LI, PA, VB, and DP wish to thank Dr. Luca Presotto for its assistance in image handling and Dr. Tommaso Ballarini and Dr. Silvia Paola Caminiti for their comments on the work.

Authors’ disclosures available online ().

References

Petersen

, Caracciolo

, Brayne

, Gauthier

, Jelic

, Fratiglioni

(2014) Mild cognitive impairment: A concept in evolution. J Intern Med 275, 214–228.

Hampel

, Frank

, Broich

, Teipel

, Katz

, Hardy

, Herholz

, Bokde

ALW

, Jessen

, Hoessler

, Sanhai

, Zetterberg

, Woodcock

, Blennow

(2010) Biomarkers for Alzheimer’s disease: Academic, industry and regulatory perspectives. Nat Rev Drug Discov 9, 560–574.

Shaw

, Korecka

, Clark

, Lee

VMY

, Trojanowski

(2007) Biomarkers of neurodegeneration for diagnosis and monitoring therapeutics. Nat Rev Drug Discov 6, 295–303.

Albert

, DeKosky

, Dickson

, Dubois

, Feldman

, Fox

, Gamst

, Holtzman

, Jagust

, Petersen

, Snyder

, Carrillo

, Thies

, Phelps

(2011) The diagnosis of mild cognitive impairment due to Alzheimer’s disease: Recommendations from the National Institute on Aging-Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease. Alzheimers Dement 7, 270–279.

McKhann

, Knopman

, Chertkow

, Hyman

, Jack

Jr , Kawas

, Klunk

, Koroshetz

, Manly

, Mayeux

, Mohs

, Morris

, Rossor

, Scheltens

, Carrillo

, Thies

, Weintraub

, Phelps

(2011) The diagnosis of dementia due to Alzheimer’s disease: Recommendations from the National Institute on Aging-Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease. Alzheimers Dement 7, 263–269.

Dubois

, Feldman

, Jacova

, Hampel

, Molinuevo

, Blennow

, DeKosky

, Gauthier

, Selkoe

, Bateman

, Cappa

, Crutch

, Engelborghs

, Frisoni

, Fox

, Galasko

, Habert

M-O

, Jicha

, Nordberg

, Pasquier

, Rabinovici

, Robert

, Rowe

, Salloway

, Sarazin

, Epelbaum

, de Souza

, Vellas

, Visser

, Schneider

, Stern

, Scheltens

, Cummings

(2014) Advancing research diagnostic criteria for Alzheimer’s disease: The IWG-2 criteria. Lancet Neurol 13, 614–629.

Dubois

, Feldman

, Jacova

, DeKosky

, Barberger-Gateau

, Cummings

, Delacourte

, Galasko

, Gauthier

, Jicha

(2007) Research criteria for the diagnosis of Alzheimer’s disease: Revising the NINCDS–ADRDA criteria. Lancet Neurol 6, 734–746.

Vos

SJB

, Verhey

, Frölich

, Kornhuber

, Wiltfang

, Maier

, Peters

, Rüther

, Nobili

, Morbelli

, Frisoni

, Drzezga

, Didic

, van Berckel

BNM

, Simmons

, Soininen

, Kłoszewska

, Mecocci

, Tsolaki

, Vellas

, Lovestone

, Muscio

, Herukka

S-K

, Salmon

, Bastin

, Wallin

, Nordlund

, de Mendonça

, Silva

, Santana

, Lemos

, Engelborghs

, Van der Mussele

, Alzheimer’s Disease NeuroImaging Initiative, Freund-Levi

, Wallin

ÅK

, Hampel

, van der Flier

, Scheltens

, Visser

(2015) Prevalence and prognosis of Alzheimer’s disease at the mild cognitive impairment stage. Brain 138, 1327–1338.

Smailagic

, Vacante

, Hyde

, Martin

, Ukoumunne

, Sachpekidis

(2015) 18F-FDG PET for the early diagnosis of Alzheimer’s disease dementia and other dementias in people with mild cognitive impairment (MCI). Cochrane Database Syst Rev 1, CD010632.

10.

Zhang

, Smailagic

, Hyde

, Noel-Storr

, Takwoingi

, McShane

, Feng

(2014) (11)C-PIB-PET for the early diagnosis of Alzheimer’s disease dementia and other dementias in people with mild cognitive impairment (MCI). Cochrane Database Syst Rev 7, CD010386.

11.

Morbelli

, Garibotto

, Van De Giessen

, Arbizu

, Chételat

, Drezgza

, Hesse

, Lammertsma

, Law

, Pappatà

, Payoux

, Pagani

, European Association of Nuclear, Medicine (2015) A Cochrane review on brain [(18)F]FDG PET in dementia: Limitations and future perspectives. Eur J Nucl Med Mol Imaging 42, 1487–1491.

12.

, Zhang

, Li

, Zheng

D-M

, Guo

, Feng

, Ren

W-D

(2014) Predictive accuracy of amyloid imaging for progression from mild cognitive impairment to Alzheimer disease with different lengths of follow-up: A meta-analysis. [Corrected]. Medicine (Baltimore) 93, e150.

13.

Frisoni

, Bocchetta

, Chételat

, Rabinovici

, de Leon

, Kaye

, Reiman

, Scheltens

, Barkhof

, Black

(2013) Imaging markers for Alzheimer disease Which vs how. Neurology 81, 487–500.

14.

Perani

, Schillaci

, Padovani

, Nobili

, Iaccarino

, Della Rosa

, Frisoni

, Caltagirone

(2014) A survey of FDG- and amyloid-PET imaging in dementia and GRADE analysis. Biomed Res Int 2014, 785039.

15.

Rowe

, Ng

, Ackermann

, Gong

, Pike

, Savage

, Cowie

, Dickinson

, Maruff

, Darby

, Smith

, Woodward

, Merory

, Tochon-Danguy

, O’Keefe

, Klunk

, Mathis

, Price

, Masters

, Villemagne

(2007) Imaging beta-amyloid burden in aging and dementia. Neurology 68, 1718–1725.

16.

Rabinovici

, Furst

, O’Neil

, Racine

, Mormino

, Baker

, Chetty

, Patel

, Pagliaro

, Klunk

, Mathis

, Rosen

, Miller

, Jagust

(2007) 11C-PIB PET imaging in Alzheimer disease and frontotemporal lobar degeneration. Neurology 68, 1205–1212.

17.

Cerami

, Della Rosa

, Magnani

, Santangelo

, Marcone

, Cappa

, Perani

(2015) Brain metabolic maps in mild cognitive impairment predict heterogeneity of progression to dementia. Neuroimage Clin 7, 187–194.

18.

Perani

, Cerami

, Caminiti

, Santangelo

, Coppi

, Ferrari

, Pinto

, Passerini

, Falini

, Iannaccone

, Cappa

, Comi

, Gianolli

, Magnani

(2016) Cross-validation of biomarkers for the early differential diagnosis and prognosis of dementia in a clinical setting. Eur J Nucl Med Mol Imaging 43, 499–508.

19.

Rabinovici

, Rosen

, Alkalay

, Kornak

, Furst

, Agarwal

, Mormino

, O’Neil

, Janabi

, Karydas

, Growdon

, Jang

, Huang

, DeArmond

, Trojanowski

, Grinberg

, Gorno-Tempini

, Seeley

, Miller

, Jagust

(2011) Amyloid vs FDG-PET in the differential diagnosis of AD and FTLD. Neurology 77, 2034–2042.

20.

Jagust

, Reed

, Mungas

, Ellis

, DeCarli

(2007) What does fluorodeoxyglucose PET imaging add to a clinical diagnosis of dementia? Neurology 69, 871–877.

21.

, Villemagne

, Masters

, Rowe

(2007) Evaluating atypical dementia syndromes using positron emission tomography with carbon 11 labeled Pittsburgh Compound B. Arch Neurol 64, 1140–1144.

22.

Della Rosa

, Cerami

, Gallivanone

, Prestia

, Caroli

, Castiglioni

, Gilardi

, Frisoni

, Friston

, Ashburner

, Perani

, Consortium EADC-PET (2014) A standardized [18F]-FDG-PET template for spatial normalization in statistical parametric mapping of dementia. Neuroinformatics 12, 575–593.

23.

Perani

, Della Rosa

, Cerami

, Gallivanone

, Fallanca

, Vanoli

, Panzacchi

, Nobili

, Pappatà

, Marcone

, Garibotto

, Castiglioni

, Magnani

, Cappa

, Gianolli

, Consortium EADC-PET (2014) Validation of an optimized SPM procedure for FDG-PET in dementia diagnosis in a clinical setting. Neuroimage Clin 6, 445–454.

24.

Nordberg

, Carter

, Rinne

, Drzezga

, Brooks

, Vandenberghe

, Perani

, Forsberg

, Langstrom

, Scheinin

, Karrasch

, Någren

, Grimmer

, Miederer

, Edison

, Okello

, Van Laere

, Nelissen

, Vandenbulcke

, Garibotto

, Almkvist

, Kalbe

, Hinz

, Herholz

(2013) A European multicentre PET study of fibrillar amyloid in Alzheimer’s disease. Eur J Nucl Med Mol Imaging 40, 104–114.

25.

Bergman

, Blomberg

, Almkvist

(2007) The importance of impaired physical health and age in normal cognitive aging. Scand J Psychol 48, 115–125.

26.

Cerami

, Crespi

, Della Rosa

, Dodich

, Marcone

, Magnani

, Coppi

, Falini

, Cappa

, Perani

(2015) Brain changes within the visuo-spatial attentional network in posterior cortical atrophy. J Alzheimers Dis 43, 385–395.

27.

Cerami

, Dodich

, Lettieri

, Iannaccone

, Magnani

, Marcone

, Gianolli

, Cappa

, Perani

(2016) Different FDG-PET metabolic patterns at single-subject level in the behavioral variant of fronto-temporal dementia. Cortex 83, 101–112.

28.

Cerami

, Dodich

, Greco

, Iannaccone

, Magnani

, Marcone

, Pelagallo

, Santangelo

, Cappa

, Perani

(2017) The role of single-subject brain metabolic patterns in the early differential diagnosis of primary progressive aphasias and in prediction of progression to dementia. J Alzheimers Dis 55, 183–197.

29.

Iaccarino

, Crespi

, Della Rosa

, Catricalà

, Guidi

, Marcone

, Tagliavini

, Magnani

, Cappa

, Perani

(2015) The semantic variant of primary progressive aphasia: Clinical and neuroimaging evidence in single subjects. PLoS One 10, e0120197–17.

30.

Presotto

, Ballarini

, Caminiti

, Bettinardi

, Gianolli

, Perani

(2017) Validation of (18)F-FDG-PET single-subject optimized SPM procedure with different PET scanners. Neuroinformatics 15, 151–163.

31.

Herholz

, Salmon

, Perani

, Baron

J-C

, Holthoff

, Frölich

, Schönknecht

, Ito

, Mielke

, Kalbe

, Zündorf

, Delbeuck

, Pelati

, Anchisi

, Fazio

, Kerrouche

, Desgranges

, Eustache

, Beuthien-Baumann

, Menzel

, Schröder

, Kato

, Arahata

, Henze

, Heiss

(2002) Discrimination between Alzheimer dementia and controls by automated analysis of multicenter FDG PET. Neuroimage 17, 302–316.

32.

Gorno-Tempini

, Hillis

, Weintraub

, Kertesz

, Mendez

, Cappa

, Ogar

, Rohrer

, Black

, Boeve

, Manes

, Dronkers

, Vandenberghe

, Rascovsky

, Patterson

, Miller

, Knopman

, Hodges

, Mesulam

, Grossman

(2011) Classification of primary progressive aphasia and its variants. Neurology 76, 1006–1014.

33.

Crutch

, Lehmann

, Schott

, Rabinovici

, Rossor

, Fox

(2012) Posterior cortical atrophy. Lancet Neurol 11, 170–178.

34.

Warren

, Fletcher

, Golden

(2012) The paradox of syndromic diversity in Alzheimer disease. Nat Rev Neurol 8, 451–464.

35.

Maldjian

, Laurienti

, Kraft

, Burdette

(2003) An automated method for neuroanatomic and cytoarchitectonic atlas-based interrogation of fMRI data sets. Neuroimage 19, 1233–1239.

36.

Tzourio-Mazoyer

, Landeau

, Papathanassiou

, Crivello

, Etard

, Delcroix

, Mazoyer

, Joliot

(2002) Automated anatomical labeling of activations in SPM using a macroscopic anatomical parcellation of the MNI MRI single-subject brain. Neuroimage 15, 273–289.

37.

Forsberg

, Engler

, Almkvist

, Blomquist

, Hagman

, Wall

, Ringheim

, Langstrom

, Nordberg

(2008) PET imaging of amyloid deposition in patients with mild cognitive impairment. Neurobiol Aging 29, 1456–1465.

38.

Kato

, Inui

, Nakamura

, Ito

(2016) Brain fluorodeoxyglucose (FDG) PET in dementia. Ageing Res Rev 30, 73–84.

39.

Heiss

W-D

, Zimmermann-Meinzingen

(2012) PET imaging in the differential diagnosis of vascular dementia. J Neurol Sci 322, 268–273.

40.

Heiss

W-D

, Rosenberg

, Thiel

, Berlot

, de Reuck

(2016) Neuroimaging in vascular cognitive impairment: A state-of-the-art review. BMC Med 14, 174.

41.

Vitali

, Migliaccio

, Agosta

, Rosen

, Geschwind

(2008) Neuroimaging in dementia. Semin Neurol 28, 467–483.

42.

Villemagne

(2016) Amyloid imaging: Past, present and future perspectives. Ageing Res Rev 30, 95–106.

43.

Jansen

, Ossenkoppele

, Knol

, Tijms

, Scheltens

, Verhey

FRJ

, Visser

, Aalten

, Aarsland

, Alcolea

, Alexander

, Almdahl

, Arnold

, Baldeiras

, Barthel

, van Berckel

BNM

, Bibeau

, Blennow

, Brooks

, van Buchem

, Camus

, Cavedo

, Chen

, Chételat

, Cohen

, Drzezga

, Engelborghs

, Fagan

, Fladby

, Fleisher

, van der Flier

, Ford

, Förster

, Fortea

, Foskett

, Frederiksen

, Freund-Levi

, Frisoni

, Froelich

, Gabryelewicz

, Gill

, Gkatzima

, Gómez-Tortosa

, Gordon

, Grimmer

, Hampel

, Hausner

, Hellwig

, Herukka

S-K

, Hildebrandt

, Ishihara

, Ivanoiu

, Jagust

, Johannsen

, Kandimalla

, Kapaki

, Klimkowicz-Mrowiec

, Klunk

, Köhler

, Koglin

, Kornhuber

, Kramberger

, Van Laere

, Landau

, Lee

, de Leon

, Lisetti

, Lleó

, Madsen

, Maier

, Marcusson

, Mattsson

, de Mendonça

, Meulenbroek

, Meyer

, Mintun

, Mok

, Molinuevo

, Møllergård

, Morris

, Mroczko

, Van der Mussele

, Na

, Newberg

, Nordberg

, Nordlund

, Novak

, Paraskevas

, Parnetti

, Perera

, Peters

, Popp

, Prabhakar

, Rabinovici

, Ramakers

IHGB

, Rami

, Resende de Oliveira

, Rinne

, Rodrigue

, Rodríguez-Rodríguez

, Roe

, Rot

, Rowe

, Rüther

, Sabri

, Sánchez-Juan

, Santana

, Sarazin

, Schröder

, Schütte

, Seo

, Soetewey

, Soininen

, Spiru

, Struyfs

, Teunissen

, Tsolaki

, Vandenberghe

, Verbeek

, Villemagne

, Vos

SJB

, van Waalwijk van Doorn

LJC

, Waldemar

, Wallin

ÅK

, Wiltfang

, Wolk

, Zboch

, Zetterberg

(2015) Prevalence of cerebral amyloid pathology in persons without dementia. JAMA 313, 1924–1915.

44.

Herrup

(2015) The case for rejecting the amyloid cascade hypothesis. Nat Neurosci 18, 794–799.

45.

Perani

(2014) FDG-PET and amyloid-PET imaging. Curr Opin Neurol 27, 405–413.

46.

Brück

, Virta

, Koivunen

, Koikkalainen

, Scheinin

, Helenius

, Någren

, Helin

, Parkkola

, Viitanen

, Rinne

(2013) [11C]PIB, [18F]FDG and MR imaging in patients with mild cognitive impairment. Eur J Nucl Med Mol Imaging 40, 1567–1572.

47.

Trzepacz

, Yu

, Sun

, Schuh

, Case

, Witte

, Hochstetler

, Hake

; Alzheimer’s Disease Neuroimaging Initiative (2014) Comparison of neuroimaging modalities for the prediction of conversion from mild cognitive impairment to Alzheimer’s dementia. Neurobiol Aging 35, 143–151.

48.

Hatashita

, Yamasaki

(2013) Diagnosed mild cognitive impairment due to Alzheimer’s disease with PET biomarkers of beta amyloid and neuronal dysfunction. PLoS One 8, e66877.

49.

Grimmer

, Wutz

, Alexopoulos

, Drzezga

, Förster

, Förstl

, Goldhardt

, Ortner

, Sorg

, Kurz

(2016) Visual versus fully automated analyses of 18F-FDG and amyloid PET for prediction of dementia due to Alzheimer disease in mild cognitive impairment. J Nucl Med 57, 204–207.

50.

Sperling

, Mormino

, Johnson

(2014) The evolution of preclinical Alzheimer’s disease: Implications for prevention trials. Neuron 84, 608–622.

A Cross-Validation of FDG- and Amyloid-PET Biomarkers in Mild Cognitive Impairment for the Risk Prediction to Dementia due to Alzheimer’s Disease in a Clinical Setting

Abstract

Keywords

INTRODUCTION

METHODS

Participants

Cognitive assessment

PET Studies

18F-FDG-PET acquisition

18F-FDG-PET data analysis

11C-PiB-PET acquisition

11C-PiB-PET data analysis

Neuroimaging biomarkers positivity

18F-FDG-PET

11C-PiB-PET

Biomarkers analysis for progression prediction

RESULTS

Cognitive assessment

Biomarkers positivity

Biomarkers progression prediction

DISCUSSION

Conclusion

Footnotes

ACKNOWLEDGMENTS

References

¹⁸F-FDG-PET acquisition

¹⁸F-FDG-PET data analysis

¹¹C-PiB-PET acquisition

¹¹C-PiB-PET data analysis

¹⁸F-FDG-PET

¹¹C-PiB-PET