Background
Several neurodegenerative disorders can lead to dementia, including Alzheimer’s disease (AD), frontotemporal dementia (FTD), dementia with Lewy bodies (DLB), vascular dementia and other less frequent disorders. To aid in the differential diagnosis of patients presenting with a cognitive disorder, the consequences of neuronal dysfunction and death can be measured by imaging.
In clinical practice, most often
18F-fluorodeoxyglucose (FDG) PET or structural magnetic resonance imaging (sMRI) (e.g. by hippocampal volume assessment in AD) are used.
18F-FDG PET is healthcare-reimbursed in several countries for differential diagnosis in cognitive dysfunction of unclear etiology, and shows high sensitivity and specificity [
1]. However,
18F-FDG PET has a disadvantage that regional glucose metabolism is a composite signal, consisting of a combination of neuronal and astrocytic activity [
2], is partly determined by functional interconnectivity with other brain areas, and can be increased by local neuroinflammation. As the latter is present in all forms of neurodegeneration, a specific measure of regional neuronal impairment may thus be masked. Furthermore, acquisition circumstances (e.g. glucose blood concentration [
3,
4]) and sensory stimulation during the tracer uptake period (e.g. light or noise), may influence the regional activity in sensory brain areas. In particular, medial temporal hypometabolism is variable and can be absent in many patients [
5‐
7].
18F-FDG is also not specific in the differential diagnosis between psychiatric disorders and neurodegeneration such as for behavioral variants of FTD [
8]. Nonetheless, some studies have shown that
18F-FDG PET correlates with dementia severity [
9,
10], and a correlation of neurocognitive parameters with regional changes in
18F-FDG uptake has been demonstrated [
11]. Interestingly, in patients with subjective cognitive impairment (showing no significant deficits on neuropsychological testing), heterogeneous patterns of hypometabolism can be observed [
12,
13] and also in cognitively normal patients with depressive symptoms AD-like hypometabolism has been reported, independently of amyloid burden [
14]. Longitudinal studies will have to further clarify the meaning of these findings. sMRI is a widely available but a less sensitive marker with also relatively low specificity [
15,
16] and quantification of hippocampal volume is not widely performed in clinical routine [
17,
18]. While it is known that brain perfusion is a less sensitive marker for neuronal activity changes in neurodegeneration compared to
18F-FDG PET, and glucose uptake impairment is an earlier biomarker [
19], brain perfusion techniques such as arterial spin labeling (ASL) are investigated as absolute regional blood flow and can be easily obtained in conjunction with other primary structural MR sequences [
20‐
22]. The reported sensitivity and specificity of ASL for discrimination of patients from healthy volunteers (HV) varies from 53 to 86% and 84–92%, respectively [
23,
24]. Differences in ASL technique and small sample sizes may be the main factors for this wide range, especially in sensitivity. Nevertheless, it has been suggested that ASL could replace the need for
18F-FDG and allow for another PET biomarker to be measured at the same time in PET/MR [
21].
Until recently, measuring synaptic density in humans required brain tissue from autopsy or surgical resection specimens. In vivo PET imaging of synaptic density has now become possible through development of radioligands with high affinity and selectivity for synaptic vesicle protein 2 A (SV2A) [
25], which is the only of three isoforms that is ubiquitously and homogeneously located/expressed in synapses across the brain.
11C-UCB-J has been thoroughly validated and is regarded as best-in-class ligand for SV2A/synaptic density PET imaging.
11C-UCB-J shows good pharmacokinetics, quantification of
11C-UCB-J distribution volume is possible and is an vivo proxy of synaptic density [
26,
27].
11C-UCB-J has been used in clinical epilepsy drug trials [
28]. Simplified quantification with a white matter reference is possible for easier use in patient populations [
29,
30]. Furthermore, stability of synaptic density in healthy aging has been shown [
31,
32]. A direct comparison in young HV, has shown differences between resting
18F-FDG uptake patterns and the regional distribution of synaptic density, with relative differences up to 30% with higher
11C-UCB-J signal especially in the hippocampus, lateral temporal cortex and cingulate, areas of particular importance in cognitive impairment [
33].
Pilot studies in patients with AD show a medial temporal decrease in synaptic density of up to 40% compared to HV [
34‐
37], while in the neocortex more moderate changes of about 10% are present whereas for
18F-FDG these are about 20% [
38]. In a direct comparison, medial temporal
11C-UCB-J signal showed strongest correlations across all cognitive domains whereas for neocortical regions,
18F-FDG uptake seems more strongly correlated [
39]. In mild cognitive impairment (MCI) there is a strong negative relationship between synaptic density and hippocampal tau accumulation [
40,
41], while in MCI and AD, SV2A binding is also highly correlated with several cognitive domain functions [
37,
40‐
42]. In FTD subtypes, which are mainly associated with aberrant TDP-43/tau/FUS protein depositions, distinct patterns of synaptic loss compared to other neurodegenerative disorders have been observed in populations including C9orf72 mutation carriers [
43], behavioral variant FTD [
44], PSP and CBD [
45,
46]. Multimodal imaging has revealed an association between lower synaptic density and reduced functional connectivity, in addition to that accounted for by grey matter atrophy [
47]. In Lewy body dementia, including Parkinson’s dementia and DLB, deposits of alpha-synuclein form the main underlying proteinopathy and compared to Alzheimer’s disease, there is often only mild atrophy. Also in this population, marked cortical synaptic loss can be observed [
48,
49].
The short half-life of
11C (20.4 min) and single tracer manufacturing per subject makes the clinical routine use of
11C-UCB-J cumbersome. Recently,
18F-SynVesT-1, an optimized
18F-labeled analogue of UCB-J with similar kinetics, binding, and test-retest properties has been evaluated in humans [
50‐
52] with also good quantitative correspondence to
11C-UCB-J.
The primary objective of this clinical trial is to determine the added value for clinical use of synaptic density imaging using a multimodal simultaneous PET/MR approach in patients with dementia and other patients developing cognitive dysfunction, by assessing the functional burden of the disease on the level of the synapse instead of synaptic activity/glucose metabolism, and by identifying an optimal combination of synaptic density and other PET/MR imaging metrics (perfusion, structural atrophy) that may allow early assessment and risk stratification for cognitive and behavioral dysfunction in de novo patients with uncertain origin of dementia. This will enable us to better understand the underlying pathophysiology of dementia, assess the direct consequences of underlying proteinopathy, relate this to subsequent structural measures and identify those parameters that can contribute to the accuracy of an early differential diagnosis. It will thereby aid the societal/economic challenge of earlier diagnosis, prognosis and biomarker development for more objective and more efficient monitoring of novel therapeutic trials. Secondary objectives are to assess how synaptic density is altered in the different cognitive disorders and how it correlates to specific symptomatology. Moreover, a direct comparison between 11C-UCB-J and 18F-SynVesT-1 in terms of distribution volume and standardized uptake value ratios (SUVR), including variability and noise levels will be conducted in HV.
Discussion
This study will compare the diagnostic accuracy of 18F-SynVesT-1 to 18F-FDG PET in cognitive disorders with uncertain etiology and in exclusion of a neurodegenerative cause in patients with cognitive impairment in late-life psychiatric disorders. The acquisition of PET and MR imaging data as well as neuropsychological testing both in patients and HV will enable us to assess not only the relationship between cognition and imaging data but also between these different imaging modalities (PET and ASL) themselves.
Using
11C-UCB-J, synaptic density has been assessed in AD, FTD and DLB, compared to HV. For AD, largest effect sizes are found in the hippocampus and other medial temporal regions extending into the posterior cingulate cortex and to a lesser extent in neocortical regions [
36], whereas in this population,
18F-FDG showed larger effect sizes especially in neocortical regions [
38]. In patients with DLB, synaptic loss was observed in substantia nigra, occipital, parietal and frontal cortices but not in medial temporal regions such as hippocampus and amygdala [
48]. In a direct comparison of
11C-UCB-J and
18F-FDG in DLB patients, the magnitude as well as spatial extent of hypometabolism exceeded that of synaptic loss [
75]. As for FTD, decreased synaptic density was most prominent in frontal regions and to a somewhat lesser extent also in temporal regions, insula and anterior cingulate [
44]. To the best of our knowledge, no direct comparison with FDG PET has been published in FTD. Although pilot studies discriminating AD from HV show larger effect sizes for hypometabolism compared to synaptic loss, regional changes e.g. in the hippocampus, may show increased specificity and effect size itself is not prohibitive to an equivalent or even better discrimination between the different dementia subtypes.
Since patients will be included upon referral for
18F-FDG PET at the UZ Leuven PET center, the sample will be highly representative for a real-life clinical setting. Another strength of our study is the use of simultaneous PET and MR imaging, which will allow us to investigate multiple modalities with limited associated burden for patients. We believe that achieving the anticipated sample size in an adequate time window is feasible due to this 1-scan study protocol in combination with the considerable extent of the local hospital memory clinic.
18F-SynVesT-1 PET will not only provide a cleaner marker of synaptic density, but it also obviates the need (i) for patients to be fasted at least four hours prior to tracer injection, (ii) to delay simultaneous MR scanning to avoid primary auditory cortex activation and (iii) to provide a dark environment prior to scanning to limit visual cortex activation. Furthermore, as visual stimulation has been shown not to change
11C-UCB-J levels in the occipital cortex [
76], no influence of scanning with eyes open or closed is expected.
As for limitations and compromises made in the design of the study, some sources of bias may be present. Expert visual readers will be more experienced in reading 18F-FDG scans compared to 18F-SynVesT-1 scans. Therefore, they will be presented a normal dataset for visual inspection before patient reads. Due to clinical need and the importance of 18F-FDG PET in the diagnostic work-up, the 18F-FDG PET result will likely influence the final clinical diagnosis and might result in a bias in favor of 18F-FDG PET. Therefore, anonymized reads will be done also by expert nuclear medicine physicians not involved in the clinical workup.
The sample size calculation was based on observed effect sizes for
11C-UCB-J as
18F-SynVesT-1 data are not yet publicly available. However, similarity of
11C-UCB-J and
18F-SynVesT-1 in terms of distribution volume and binding potential as well as test-retest characteristics has been demonstrated [
51,
52]. Accordingly, it can be expected to find similar effect sizes for
18F-SynVesT-1 as for
11C-UCB-J and we believe the anticipated sample size will result in adequate power. Another limitation will be the short acquisition protocol for the patient group and the associated simplified quantification method (SUVR), which was a trade-off between scanning a larger sample with simplified quantification or a smaller sample with full dynamic quantification. Of note, performing 90-minute dynamic acquisitions in a clinical setting is not feasible. As we aim to investigate the clinical applicability of
18F-SynVesT-1, the use of short static acquisitions is justified. The single-tracer nature of the study can be seen as a limitation since we cannot for example determine correlations with tau or amyloid PET imaging data. Nevertheless,
18F-FDG PET data will be acquired for clinical work-up in all patients (as inclusion criteria) and most patients will also receive lumbar punctions to determine cerebrospinal fluid levels of Aβ42/Aβ40 and total tau to exclude or confirm AD. Therefore, performing retrospective (sub)analyses using these data might also be possible upon study completion and ethics approval.
In conclusion, this study will provide further insight into synaptic density PET and its diagnostic applicability in a clinical routine setting.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.