Quantification of 18 F-SynVesT-1 binding in healthy ageing and head-to-head comparison with 11 C-UCB-J

Abstract

Synaptic vesicle glycoprotein 2A (SV2A) PET imaging enables the estimation of synaptic density in the brain. The most widely used SV2A tracer, ¹¹C-UCB-J, shows age-related effects in healthy volunteers (HV) and its short radioactive half-life limits clinical applicability. ¹⁸F-SynVesT-1 has a longer half-life, but its quantification and dependence on age/sex need further validation. In this study, 40 HV aged 24–85 (21F/19M) underwent 90-min ¹⁸F-SynVesT-1 PET–MR with arterial sampling. Distribution volumes (V_T) and distribution volume ratios (DVR) were estimated using one-tissue (1TC) and two-tissue (2TC) compartment models, and Logan graphical analysis (LGA). 1TC and LGA showed strong agreement with 2TC V_T (R² ⩾ 0.97). Standardised uptake value ratios (SUVR) were evaluated across different time windows, using the centrum semiovale (CSO) and cerebellar (pseudo-)reference regions. SUVR_CSO 60–90 min correlated best with DVR_CSO. The thalamus and cingulate cortex showed age-related decreases in ¹⁸F-SynVesT-1 DVR_CSO (~2%/decade). Ten subjects (52 ± 21 years; 5F/5M) also underwent ¹¹C-UCB-J PET for a head-to-head comparison. ¹⁸F-SynVesT-1 showed lower cortical V_T (−7% ± 9%) but higher DVR_CSO (37% ± 4%). This study shows that ¹⁸F-SynVesT-1 can be reliably quantified by SUVR in healthy volunteers, and correlates well with ¹¹C-UCB-J. These findings support the broader application of ¹⁸F-SynVesT-1. This study was registered on clinicaltrial.gov (NCT05384353).

Keywords

Ageing kinetic modelling SV2A SynVesT-1 UCB-J

Introduction

Synaptic vesicle glycoprotein 2A (SV2A) is a presynaptic protein, that is, expressed in most synapses of the human brain.¹ Positron emission tomography (PET) imaging of SV2A has been regarded as a proxy for measuring synaptic density in vivo.² The most widely published SV2A PET tracer is ¹¹C-UCB-J.^{3
¹¹}C-UCB-J binding was shown to be negatively correlated with age in the caudate nucleus but not influenced by sex.^{4
–6 ¹¹}C-UCB-J imaging studies have demonstrated lower SV2A binding in multiple neurodegenerative disorders⁷ such as Alzheimer’s,^8,9 Parkinson’s¹⁰ and Huntington’s disease¹¹ and showed that tracer binding was significantly associated with clinical symptoms.^12
–14 However, the short radioactive half-life of ¹¹C-UCB-J (about 20 min) is suboptimal for widespread research and potential clinical use. To overcome this, a UCB-J derivative labelled with ¹⁸F, ¹⁸F-SynVesT-1, was developed.¹⁵ A first-in-human evaluation in eight healthy volunteers (HV) demonstrated that ¹⁸F-SynVesT-1 displays excellent kinetics and binding properties in the human brain and that its regional distribution volumes (V_T) and binding potentials (BP_ND) are comparable to those of ¹¹C-UCB-J.¹⁵ In this pilot study, a one tissue compartment model was found to fit the kinetic profile of ¹⁸F-SynVesT-1 best. In a subsequently published subset of seven HV, a two-parameter variant of a simplified reference tissue model (SRTM2) that uses a global value for the washout time constant in the reference region was found to produce little bias for calculation of BP_ND.¹⁶ However, both studies included a low number of participants and to the best of our knowledge, a validation in an independent and larger HV cohort has not been performed. Moreover, the influence of age and sex on ¹⁸F-SynVesT-1 binding has not been investigated.

This study included 40 HV (aged 24–85) who underwent dynamic ¹⁸F-SynVesT-1 PET-MR scanning with full kinetic modelling. First, we validated which kinetic model was most suitable for quantification of ¹⁸F-SynVesT-1 and whether a Logan graphical analysis (LGA) would result in an appropriate fit of the data as well. Second, we assessed which SUVR time windows and (pseudo-)reference regions most closely matched full kinetic quantification. Third, we investigated the association of ¹⁸F-SynVesT-1 binding with age and sex. Finally, we directly compared ¹⁸F-SynVesT-1 and ¹¹C-UCB-J in a subset of 10 HV with regard to V_T and kinetic microparameters.

Methods

Participants

Adult HV aged 18–85 were recruited through local newspapers and internet advertisements between June 2022 and January 2025. Inclusion and exclusion criteria are given in the Supplementary Material. In short, participants had no history of any major internal, neurological or psychiatric disease, no history of substance abuse and they were not a first-degree relative of a person with a neurological or psychiatric history.

All participants with mini-mental state examination (MMSE) scores lower than 28 were excluded and participants aged >50 underwent amyloid PET imaging with ¹⁸F-NAV4694 for subcohort characterisation. These images were visually interpreted by an experienced reader (KVL) in a binary fashion to determine amyloid positivity.

The study was approved by the ethics committee UZ/KU Leuven and was conducted according to the latest version of the Declaration of Helsinki. The study was registered on clinicaltrials.gov (NCT05384353). All participants gave written informed consent prior to study enrolment.

PET data acquisition and processing

All HVs underwent a 90-min ¹⁸F-SynVesT-1 PET-MR (mean injected activity: 145 ± 12 MBq; molar activity: 332 ± 191 GBq/μmol; injected mass dose: 0.27 ± 0.17 μg). Ten participants additionally underwent a 90-min ¹¹C-UCB-J PET-MR (mean injected activity: 262 ± 60 MBq; molar activity: 340 ± 167 GBq/μmol; injected mass dose: 0.62 ± 0.34 μg). ¹⁸F-SynVesT-1 and ¹¹C-UCB-J were administered as a bolus injection. Arterial blood samples were drawn every 10 s from 10 to 90 s, every 15 s from 90 s to 3 min; and then at 3.5, 5, 6.5, 8, 12, 15, 20, 25, 30, 45, 60, 75 and 90 min for both tracers. Arterial blood and plasma samples were counted with a calibrated well counter. Additional arterial blood samples were collected at 3, 8, 15, 30, 60 and 90 min for metabolite analysis.

All PET images were acquired using a 3T General Electric (GE) Signa PET-MR system with time-of flight PET acquisition. Dynamic scan data were reconstructed in 30 frames (6 × 15 s, 3 × 30 s, 3 × 60 s, 3 × 3 min, 15 × 5 min) with corrections for attenuation, normalisation, scatter, deadtime and radioactive decay. Simultaneously with the PET acquisition, a 3D T1-weighted MRI was acquired using a 32-channel head coil. Standardised uptake values (SUV) were generated, and motion correction was performed by rigid co-registration of all frames to an average of the early frames (PMOD v4.1; Zurich, Switzerland). Subsequently, PET images were co-registered to the corresponding T1-weighted MRI (PMOD v4.1) and segmented with the CAT12 toolbox of Statistical Parametric Mapping (SPM12; Wellcome Trust Centre for Neuroimaging). PET volumes of interest (VOIs) were defined based on tissue probability maps, the Neuromorphometrics atlas and the Mori atlas (for the pons). A centrum semiovale (CSO) reference region VOI was determined as previously described.¹⁷ The following 12 grey matter regions were selected for further analysis: frontal, parietal, occipital and temporal cortex, cingulate (anterior and posterior combined), insula, hippocampus, amygdala, putamen, caudate, thalamus and cerebellar (CBL) cortex. Regional tissue time-activity curves (TACs) were generated and weighted for decay and frame duration for fitting (PMOD v4.1). Whole blood and plasma curves were linearly interpolated. Metabolite fractions (intact parent tracer) were fitted using a sigmoid curve. For ¹⁸F-SynVesT-1, regional V_T values were computed using one and two-tissue compartment (1TC, 2TC) models and LGA. For LGA, a t* of 30 min post injection and fixed vB of 0.05 were used. For ¹¹C-UCB-J, a 1TC model was used as previously validated.^18,19 Additionally, parametric V_T images were created using LGA and were used for further analyses including distribution volume ratios (DVR). DVR was calculated as V_T (target region)/V_T (reference region). The CSO was used as the main validated reference region to compute regional DVR. Since the CSO is not completely devoid of specific binding, we regarded this as a pseudo-reference region and therefore BP_ND was not calculated. Exploratively, the pons and three different cerebellar regions (CBL white matter, CBL cortex and whole CBL) were tested as potential pseudo-reference regions to assess their validity when using an SUVR simplification. Mean SUVR were calculated using 15- and 30-min windows across each reference region and assessed at timepoints in 5-min increments post-injection, in comparison to DVR values derived from the parametric V_T images. Partial volume correction (PVC) was applied to SUVR and parametric LGA images using a region-based voxelwise technique.²⁰

Statistical analysis

Statistical analyses were performed in RStudio 4.3.2 (R Foundation for Statistical Computing, Vienna, Austria) and Python 3.11.

F-tests were performed with 1TC as the nested model for all fits (12 regions for 40 subjects = 480 fits). Additionally, Akaike information criterion (AIC) values were compared between 1TC and 2TC models. To compare micro- and macroparameters between kinetic models, a Wilcoxon signed rank test was used, because of the outliers that can occur in these comparisons (e.g. very high percentage standard error (%SE) for individual fits).

Results and differences between parameters were expressed in mean or median depending on the normality distribution of the underlying data. Coefficients of variation (CoV) were calculated by dividing the standard deviation by the mean of the DVR.

Linear regression models were used to compare kinetic model parameters of different models and both tracers and to assess the relationship with age. Next to visual assessment of scatter and residual plots, normality and homoscedasticity were formally assessed by Shapiro-Wilk tests and Bruesch–Pagan tests. Outlier V_T values (determined as values ⩾50) were excluded from these regression analyses. For the comparison of parametric V_T images obtained using LGA and SUVR-derived estimates from different time windows, voxelwise linear regressions were performed separately for each SUVR time window. Each regression included all voxels within subject specific grey matter masks, obtained by thresholding the grey matter probability map at 0.5. These were taken together in a single population-level fit between SUVR and DVR. Based on this global fit, residual values were calculated for every voxel in each subject, and these residuals were then averaged across subjects to generate parametric residual maps reflecting regional deviations from the overall population relationship.

VOI-based analyses were considered significant below a p value of 0.0042 (Bonferroni corrected for 12 regions).

Results

Demographics

We included a total of 40 HVs. The median age was 60 years (range 24–85), and the cohort included 19 males. The mean MMSE score was 30 ± 0.6; four subjects (age 53–77, 3M/1F) were cognitively unimpaired but positive for amyloid-β pathology on ¹⁸F-NAV4694 PET. For the 10 subjects who underwent both ¹⁸F-SynVesT-1 and ¹¹C-UCB-J (age 58 ± 19; 5M/5F), the average time between both scans was 12 ± 5 weeks.

¹⁸F-SynVesT-1 kinetic modelling

A 2TC model resulted in better fits with lower AIC values compared to a 1TC model in all regions except for the amygdala, putamen and caudate. Examples of fitted TACs for frontal cortex, hippocampus, cerebellum, putamen and CSO are shown in Figure 1. F-tests favoured a 2TC model in 72% of fits. While a 1TC model could reliably estimate V_T, K₁ and k₂ with %SE <10% in all regions and subjects, a 2TC model sporadically failed to fit V_T and K₁, yet still with %SE <10% in 98% (V_T) and 96% (K₁) of regions. Moreover, k₂, k₃ and k₄ could not be reliably estimated by the 2TC model. The %SE of V_T,2TC in the CSO was >10% in 45% of cases, while V_T,1TC was always reliably estimated (%SE ⩽3%), only allowing for reliable DVR_CSO calculations with 1TC.

Figure 1.

Fitted time activity curves with one- and two-tissue compartment model from one representative subject for ¹⁸F-SynVesT-1 and ¹¹C-UCB-J.

V_T,1TC was highly correlated to V_T,2TC (slope 0.99, R² = 0.97; outlier values (1.5% of fits) were excluded from this analysis; Supplementary Figure 1(A)). The 1TC model showed a modest underestimation of K₁ compared to the 2TC model (slope 1.09, R² = 0.91; no excluded outliers; Supplementary Figure 1(B)). Mean ¹⁸F-SynVesT-1 values of V_T,1TC, DVR_CSO,1TC, K_1,1TC and k_2,1TC for each VOI for the 40 HV can be found in Supplementary Table 1.

V_T,LGA values were overall slightly higher compared to V_T,1TC (median percent increase ranging from 0.1% (putamen) to 4% (cerebellar cortex), p < 0.004 in 10 out of 12 VOIs), and showed comparable %SE in most regions. Regression analysis confirmed that V_T,LGA could reliably estimate V_T,1TC values (slope 0.99, R² = 0.99). Also, V_T,2TC correlated very well with V_T,LGA (slope 0.99, R² = 0.98; after excluding seven outlier values; Supplementary Figure 2). Because of its robustness, the LGA method was used to calculate parametric images for V_T,LGA and DVR_CSO,LGA. Given the strong agreement with V_T,2TC and the more robust estimation of microparameters obtained via the 1TC, we opted to use 1TC and LGA for subsequent analyses.

Time stability of ¹⁸F-SynVesT-1 SUVR

We compared SUVR_CSO at different time windows (15- and 30-min intervals) with DVR_CSO. A voxelwise linear regression within a whole brain grey matter mask was performed for each of these time windows. Parametric images of the regression residuals were computed to investigate the spatial distribution of the variance and to identify potential regional kinetic differences.

The SUVR_CSO time window with the least bias based on slope was around 50–65 min (slope = 0.99). Earlier SUVR_CSO time windows had slopes above 1 (underestimation of high-binding regions compared to DVR_CSO) and later windows had slopes below 1 (overestimation of high-binding regions). The latest SUVR_CSO time window of 60–90 min p.i., that was often used in previous studies,^15,21,22 had the lowest variance with an R² of 0.93 within the whole grey matter VOI (Supplementary Table 2).

Parametric residual images (Figure 2) showed that the medial frontal cortex, temporal cortex, anterior cingulate cortex, insula, precuneus and amygdala had relatively high positive residual scores, especially in the early time windows. Regions with negative residual scores included mainly the cerebellar cortex, the lateral frontal cortex and the caudate. Of note, the thalamus had relatively high residual scores, negative in the early time windows and positive in the later time windows. Especially around time window 50–80 min, there were different residuals within the thalamus (Figure 2). When residuals were calculated based on a perfect fit (y = x), it became apparent that DVR_CSO in the anterolateral part of the thalamus was increasingly underestimated by SUVR_CSO at later time windows (Supplementary Figure 3).

Figure 2.

Parametric residual images comparing ¹⁸F-SynVesT-1 SUVR_CSO (x) to DVR_CSO (y) on a voxel-level at different timeframes post-injection. Blue regions represent negative residuals (overestimation of DVR_CSO by SUVR_CSO using the provided fit). A 3 mm gaussian smoothing was applied. Residuals were calculated within a grey matter mask.

Comparison of simplified SUVR measures using various reference regions

Since the CSO region is variable depending on WM and cortical anatomy and prone to noise effects, CoV for the DVR outcome variables for CBL and pons as (pseudo)reference region were also calculated. CoV were slightly higher for DVR_CSO and DVR_PNS compared to DVR_CBL (Supplementary Table 3).

A comparison of SUVR with DVR_LGA images for each reference region using a voxelwise linear regression within a whole grey matter mask was performed for each of the evaluated SUVR time windows (30-min intervals). Also here, the latest time windows always yielded the highest R² (Supplementary Table 4). Using the 60–90 min time window, the whole CBL (slope 1.03, R² = 0.98) and CBL cortex (slope 1.03, R² = 0.98) yielded almost identical results. CBL white matter gave a slope of 0.97, R² = 0.96. Using the pons as reference region resulted in a slope of 0.82, R² = 0.92. Voxelwise parametric residual images had similar variance patterns as for DVR_CSO.

¹⁸F-SynVesT-1 binding in relation to age and sex

Using parametric LGA images with PVC, linear regressions showed non-significant correlations of ¹⁸F-SynVesT-1 V_T,LGA values with age in the cingulate cortex (slope −0.09; R² = 0.14; p = 0.016) and thalamus (slope −0.06; R² = 0.14, p = 0.017), corresponding to a reduction per decade (RPD) of 2.5% (mean V_T = 30.5) and 2.7% (mean V_T = 17.6), respectively. Correlation plots are shown in Supplementary Figure 4. Significant correlations were found between age and DVR_CSO in the cingulate cortex (RPD of 2.0%; p = 0.004) and thalamus (2.2%; p = 0.0002).

No sex differences were found for V_T,1TC or DVR_CSO with or without PVC. Unpaired t-tests showed systematically higher K_1,1TC (±10%; p = 0.01–0.16, depending on the region) and k_2,1TC (±12%; p ⩽ 0.03) values in grey matter regions in females compared to males. This was not the case for microparameters in white matter regions. More detailed results on ¹⁸F-SynVesT-1 binding in relation to age and sex are shown in Supplementary Tables 5 to 7.

¹⁸F-SynVesT-1 head-to-head comparison to ¹¹C-UCB-J

Using a 1TC model, reliable estimates of V_T,1TC, K_1,1TC and k_2,1TC were obtained for both ¹⁸F-SynVesT-1 and ¹¹C-UCB-J. V_T,1TC values in a composite cortical VOI for ¹⁸F-SynVesT-1 were 7% ± 9% lower than those for ¹¹C-UCB-J (p = 0.02), while V_T,1TC in CSO was 32% ± 6% lower (p < 0.001). Although K_1,1TC did not differ significantly between the tracers, ¹⁸F-SynVesT-1 showed higher k_2,1TC values in the cerebral cortex (13% ± 16%; p = 0.03) and in the CSO (57% ± 22%; p < 0.001). A detailed summary of regional results is provided in Table 1.

Table 1.

Head-to-head comparison of SYNV1 and UCB-J kinetic parameters using a 1TC model (mean ± SD, n = 10). No partial volume correction was performed.

Region	SYNV1, DVR_CSO	UCB-J, DVR_CSO	SYNV1, V_T	UCB-J, V_T	SYNV1, K₁ (×10⁻¹)	UCB-J, K₁ (×10⁻¹)	SYNV1, k₂ (×10⁻²)	UCB-J, k₂ (×10⁻²)
Frontal Ctx	6.1 ± 0.2	4.5 ± 0.2	20.5 ± 1.6	22.2 ± 1.7	4.3 ± 0.6	4.1 ± 0.5	2.1 ± 0.4	1.9 ± 0.3
Temporal Ctx	6.6 ± 0.3	4.8 ± 0.3	22.1 ± 2.0	23.8 ± 2.0	3.7 ± 0.6	3.6 ± 0.5	1.7 ± 0.3	1.5 ± 0.2
Parietal Ctx	6.5 ± 0.4	4.8 ± 0.3	21.8 ± 1.6	23.7 ± 2.0	4.4 ± 0.6	4.2 ± 0.5	2.0 ± 0.3	1.8 ± 0.2
Occipital Ctx	6.3 ± 0.5	4.6 ± 0.4	21.2 ± 2.0	22.7 ± 1.8	4.2 ± 0.6	4.0 ± 0.6	2.0 ± 0.3	1.8 ± 0.2
Insular Ctx	6.8 ± 0.4	5.1 ± 0.3	22.9 ± 2.1	25.0 ± 2.0	4.0 ± 0.7	3.8 ± 0.6	1.8 ± 0.4	1.5 ± 0.2
Cingulate Ctx	6.6 ± 0.4	4.9 ± 0.3	22.2 ± 2.0	24.4 ± 2.4	4.3 ± 0.8	4.1 ± 0.6	2.0 ± 0.3	1.7 ± 0.2
Hippocampus	5.1 ± 0.3	3.9 ± 0.3	17.3 ± 1.7	19.2 ± 1.4	3.3 ± 0.5	3.2 ± 0.4	1.9 ± 0.3	1.7 ± 0.2
Amygdala	6.6 ± 0.4	4.9 ± 0.4	22.2 ± 2.5	24.2 ± 2.3	3.2 ± 0.4	3.0 ± 0.4	1.4 ± 0.2	1.3 ± 0.2
Thalamus	4.7 ± 0.3	3.6 ± 0.2	15.7 ± 1.6	17.6 ± 1.3	4.8 ± 0.6	4.6 ± 0.5	3.1 ± 0.4	2.6 ± 0.3
Putamen	7.7 ± 0.5	5.8 ± 0.4	25.9 ± 2.5	28.6 ± 2.1	5.4 ± 0.9	5.1 ± 0.7	2.1 ± 0.5	1.8 ± 0.3
Caudate	6.5 ± 0.4	4.9 ± 0.4	21.9 ± 1.9	24.1 ± 1.8	4.5 ± 0.8	4.3 ± 0.7	2.1 ± 0.4	1.8 ± 0.3
Whole cerebellum	4.7 ± 0.4	3.5 ± 0.3	15.9 ± 1.9	17.1 ± 1.2	3.9 ± 0.6	3.8 ± 0.5	2.5 ± 0.3	2.2 ± 0.2
Cerebellar Ctx	5.2 ± 0.5	3.8 ± 0.3	17.5 ± 2.1	18.5 ± 1.5	4.0 ± 0.6	3.9 ± 0.5	2.3 ± 0.3	2.1 ± 0.2
Cerebellar WM	2.5 ± 0.2	2.0 ± 0.2	8.4 ± 0.9	10.1 ± 0.6	3.1 ± 0.4	3.1 ± 0.4	3.7 ± 0.3	3.0 ± 0.3
CSO	1.0 ± 0.0	1.0 ± 0.0	3.4 ± 0.2	5.0 ± 0.4	1.3 ± 0.2	1.2 ± 0.2	3.9 ± 0.6	2.5 ± 0.4
Pons	2.2 ± 0.1	1.8 ± 0.1	7.4 ± 0.7	8.9 ± 0.6	3.9 ± 0.6	3.6 ± 0.4	5.3 ± 0.7	4.1 ± 0.4

Ctx: cortex, WM: white matter, CSO: centrum semiovale; SYNV1: ¹⁸F-SynVesT-1, UCB-J: ¹¹C-UCB-J.

Reference regions indicated in italic.

Consequently, cortical DVR_CSO values derived from the 1TC model were on average 37% ± 4% higher with ¹⁸F-SynVesT-1 compared to ¹¹C-UCB-J (p < 0.001; Figure 3). Notably, the cortical CoV in DVR_CSO was 7% lower for ¹⁸F-SynVesT-1 compared to ¹¹C-UCB-J in this paired cohort, mainly determined by the frontal cortex (−19%), hippocampus (−22%), caudate (−19%) and amygdala (−14%). However, for other regions, ¹⁸F-SynVesT-1 had a higher CoV (e.g. thalamus (+10%), putamen (+6%), parietal cortex (+6%) and cerebellar cortex (+5%)).

Figure 3.

Mean DVR_CSO parametric image of 10 healthy subjects that underwent both ¹⁸F-SynVesT-1 and ¹¹C-UCB-J PET.

Regression analysis showed that V_T,1TC and DVR_CSO of ¹⁸F-SynVesT-1 were strongly correlated with these respective parameters for ¹¹C-UCB-J, with a slope of 0.88 (R² = 0.71) for V_T,1TC and a slope of 0.72 (R² = 0.93) for DVR_CSO (Figure 4). Regression analyses for DVR_CSO for each individual VOI showed consistent findings (Supplementary Table 8 and Supplementary Figure 5).

Figure 4.

Linear regression for 1TC V_T and DVR_CSO values between ¹⁸F-SynVesT-1 and ¹¹C-UCB-J. Each data point represents one region of one subject.

Discussion

We evaluated the kinetic properties of ¹⁸F-SynVesT-1 binding and simplified quantification measures in a cohort of 40 HVs and compared kinetic and binding parameters to ¹¹C-UCB-J in a subcohort of 10 HVs. Consistent with prior studies, the 2TC model emerged as the preferred approach for ¹⁸F-SynVesT-1 when only taking AIC and F-tests into account. However, 2TC only estimated the V_T and K₁ parameters reliably with occasional outliers in 3%–4% of fits and could not reliably estimate k₂, k₃ and k₄. In line with Naganawa et al., we conclude that the 1TC model is the most useful model for quantification of ¹⁸F-SynVesT-1 data.¹⁵ K_1,1TC estimates were slightly lower than K_1,2TC which may represent a slight underestimation of K₁ by the 1TC model due to a lower fit accuracy in the presence of two kinetically distinct tissue compartments (fast exchange and slower binding). This can also be seen from model fits (Figure 1), where the 1TC model systematically slightly underfits the early upward slope of the tissue time–activity curves.

As for simplified quantification of ¹⁸F-SynVesT-1, DVR_CSO could be reliably approached by SUVR_CSO at later time windows from 55 min p.i. onwards. The early time windows tended to underestimate DVR_CSO, whereas later time windows accounted for the variance more effectively (R² up to 0.93) but generally overestimated the binding, especially in high-binding regions. This is consistent with a WM reference region with relatively fast kinetic properties compared to cortical GM.

Interestingly, parametric maps indicated that the anterolateral thalamus region showed increasing underestimation of DVR_CSO by SUVR_CSO at later time windows (Supplementary Figure 3). This pattern suggests that this particular thalamic subdivision may exhibit faster kinetic properties compared to the WM CSO and that the CSO, even though mostly devoid of SV2A,¹⁵ is not entirely representative of the non-specific binding throughout the entire brain. This was also reported for ¹¹C-UCB-J by Rossano et al.,²³ who calculated a parametric image of the non-displaceable uptake (V_ND) displaying substantial heterogeneity across the brain. They reported that CSO V_T exceeded the V_ND in the grey matter but did not specify regional details on the thalamus or its subregions. Future ¹⁸F-SynVesT-1 blocking studies on high-resolution scanners will have to elucidate our finding.

The CSO was previously validated as the region with the least amount of specific binding and thus has been used as a reference region for SUVR calculation. However, it is a small region, sentitive to noise, partial volume effects, white matter hyperintensities and motion. Therefore, we explored SUVR simplification using four different pseudo-reference regions. The different cerebellar regions seem to provide very similar results, while the DVR_pns could be estimated less adequately using SUVR. Furthermore, the results indicate that the intersubject variability is lower when using cerebellar pseudo-reference regions compared to pons and also compared to CSO. Cerebellar pseudo-reference regions may thus be used for calculation of SUVR with minimal bias compared to DVR²⁴ in HV but might not be a suitable pseudo-reference region in all patient populations.

After PVC, age correlations with DVR_CSO in the thalamus and cingulate cortex, corresponding to a RPD of around 2% remained significant after correction for multiple comparisons. This is in contrast with previous studies investigating the age-dependency of ¹¹C-UCB-J, where the caudate nucleus showed a significant decrease in synaptic density with increasing age.^4
–6 To the best of our knowledge, the correlation of ¹⁸F-SynVesT-1 binding with age has not been studied previously. Long-term follow-up in an independent cohort will be needed to corroborate this finding.

We found no sex difference in V_T or DVR, although females had faster kinetics than males in grey matter regions. This could be consistent with a higher cerebral blood flow in females as reported previously.²⁵ Of note, the previous studies investigating microparameters for ¹⁸F-SynVesT-1 were performed in male-dominated cohorts.^15,16

In contrast to the findings of Naganawa et al.,¹⁶ our k₂ value in the CSO was relatively high for ¹⁸F-SynVesT-1 compared to the other regions of interest. This could be due to differences in the definition of the CSO. For example, the volume of our CSO template is 2.9 mL which resulted in this group into a minimal volume of 1.5 mL in patient space. The CSO template used by Naganawa et al.¹⁶ was reported to have an initial volume of 2 mL, leading to even smaller volumes in patient space and thereby potentially increasing the sensitivity to noise. Moreover, this could potentially also be due to differences in attenuation and scatter correction between the different scanners. Taken together, our results support the use of the CSO as a reference region with faster washout kinetics.

Lastly, this study also confirmed that V_T values for ¹⁸F-SynVesT-1 were slightly lower than those observed for ¹¹C-UCB-J, which is mainly due to a higher k₂ value for ¹⁸F-SynVesT-1. Because of lower V_T,CSO for ¹⁸F-SynVesT-1, DVR_CSO were 34%–42% higher for ¹⁸F-SynVesT-1 compared to ¹¹C-UCB-J. Naganawa et al. also showed that ¹⁸F-SynVesT-1 V_T values were lower than those for ¹¹C-UCB-J while ¹⁸F-SynVesT-1 showed higher BP_ND.¹⁵ The differences across the mean BP_ND values for all assessed VOIs as reported by Naganawa et al. are ranging from 14% to 24% which is slightly lower than the difference in DVR_CSO we report here. Differences in reference regions and scanners could potentially explain this discrepancy. Nonetheless, ¹⁸F-SynVesT-1 DVR_CSO values were well correlated to ¹¹C-UCB-J DVR_CSO (slope 0.72, R² = 0.93). These findings suggest that both tracers bind specifically to the same target, with ¹⁸F-SynVesT-1 exhibiting higher affinity.¹⁵ The k₂ value in CSO was also notably higher for ¹⁸F-SynVesT-1 than for ¹¹C-UCB-J. This can be explained by the slightly lower lipophilicity of ¹⁸F-SynVesT-1 and a suggested lower affinity to the non-displaceable component.²⁶ The spatial patterns of residual differences between SUVR_CSO and DVR_CSO were similar between ¹⁸F-SynVesT-1 and ¹¹C-UCB-J (Supplementary Figure 6). The CoV of DVR_CSO in a composite cortical VOI was 7% lower for ¹⁸F-SynVesT-1 compared to ¹¹C-UCB-J. This might indicate that ¹⁸F-SynVesT-1 can show higher sensitivity to detect SV2A changes compared to ¹¹C-UCB-J.

Limitations of our study include the absence of patient data, which restricts these findings to HV. Moreover, our analyses did not account for plasma protein binding, which was shown to increase with age and to influence ¹¹C-UCB-J V_T values.⁶ The resolution of the PET system that was used in this study is lower than that of the High Resolution Research Tomograph which was used in the previous study.¹⁵ This lower resolution may introduce more partial volume effect that could influence the quantification especially in the CSO. To further investigate this as well as to assess the distinct kinetic properties and age-related changes within the thalamus, high-resolution PET systems^27,28 should be considered.

To conclude, ¹⁸F-SynVesT-1 PET allows reliable quantification of SV2A binding as proxy for synaptic density in healthy individuals. Simplified quantification using SUVR methods is most robust using later timeframes such as 60–90 min post injection. The kinetic profile and binding parameters of ¹⁸F-SynVesT-1 show strong correlation with ¹¹C-UCB-J. This study demonstrates that ¹⁸F-SynVesT-1 is a clinically useful alternative for SV2A imaging with excellent kinetic properties.

Supplemental Material

sj-docx-1-jcb-10.1177_0271678X261454164 – Supplemental material for Quantification of 18F-SynVesT-1 binding in healthy ageing and head-to-head comparison with 11C-UCB-J

Supplemental material, sj-docx-1-jcb-10.1177_0271678X261454164 for Quantification of 18F-SynVesT-1 binding in healthy ageing and head-to-head comparison with 11C-UCB-J by Greet Vanderlinden, Charles Carron, Jolien Van Opstal, Marie Cohilis, Wim Vandenberghe, Michel Koole and Koen Van Laere in Journal of Cerebral Blood Flow & Metabolism

Footnotes

Acknowledgements

We thank the research technologists Kwinten Porters, Jef Van Loock and Nele Eecloo for data acquisition, the PET radiopharmacy team of UZ Leuven for their skilled contributions and all volunteers for their participation in this study.

Author contributions

KVL and GV contributed to study design. KVL supervised study progression and obtained funding. GV performed data acquisition. GV, CC, JVO and MC performed data processing. MK, KVL, GV, CC and JVO contributed to interpretation of the results. GV, CC and JVO drafted the manuscript. MK, KVL, MC and WV critically reviewed the manuscript.

Funding

The authors disclosed receipt of the following financial support for the research, authorship and/or publication of this article: This study was supported by a grant from the Belgian Alzheimer Research foundation (SAO-FRA 2021/0004). GV and JVO are supported by a Research Foundation Flanders (FWO) grant (1292326N and 1S06625N, respectively).

Declaration of conflicting interests

The authors declared the following potential conflicts of interest with respect to the research, authorship and/or publication of this article: MK and KVL have performed contract research through KU Leuven for Merck, Johnson & Johnson, UCB, Syndesi, Eikonizo, GE Healthcare, Cerevel, BMS and Curasen. KVL is an advisory board member of Enigma.

Ethical considerations

The study was approved by the ethics committee UZ/KU Leuven (S66111) and was conducted according to the latest version of the Declaration of Helsinki.

Consent to participate

All participants gave written informed consent prior to study enrolment.

ORCID iD

Greet Vanderlinden

Data availability statement

Anonymised data will be deposited in an access-controlled file server used by the academic research PET imaging group, which will be shared on reasonable request from any qualified investigator on approval by the Ethics Committee of the local university hospital.

Supplemental material

Supplemental material for this article is available online.

References

Bajjalieh

Frantz

Weimann

, et al. Differential expression of synaptic vesicle protein 2 (SV2) isoforms. J Neurosci 1994; 14(9): 5223–5235.

Finnema

Nabulsi

Eid

, et al. Imaging synaptic density in the living human brain. Sci Transl Med 2016; 8: 348ra96.

Nabulsi

Mercier

Holden

, et al. Synthesis and preclinical evaluation of ¹¹C-UCB-J as a PET tracer for imaging the synaptic vesicle glycoprotein 2A in the brain. J Nucl Med 2016; 57: 777–784.

Michiels

Delva

Van Aalst

, et al. Synaptic density in healthy human aging is not influenced by age or sex: a ¹¹C-UCB-J PET study. Neuroimage 2021; 232: 117877.

Andersen

Hansen

Knudsen

, et al. Healthy brain aging assessed with [¹⁸F]FDG and [¹¹C]UCB-J PET. Nucl Med Biol 2022; 112–113: 52–58.

Toyonaga

Khattar

, et al. The regional pattern of age-related synaptic loss in the human brain differs from gray matter volume loss: in vivo PET measurement with [¹¹C]UCB-J. Eur J Nucl Med Mol Imaging 2024; 51: 1012–1022.

Carson

Naganawa

Toyonaga

, et al. Imaging of synaptic density in neurodegenerative disorders. J Nucl Med 2022; 63: 60S–67S.

Vanderlinden

Koole

Michiels

, et al. Longitudinal synaptic loss versus tau Braak staging in amnestic mild cognitive impairment. Alzheimers Dement 2025; 21: e14412.

Mecca

Chen

O’Dell

, et al. In vivo measurement of widespread synaptic loss in Alzheimer’s disease with SV2A PET. Alzheimers Dement 2020; 16(7): 974–982.

10.

Delva

Van Laere

Vandenberghe

Longitudinal positron emission tomography imaging of presynaptic terminals in early Parkinson’s disease. Mov Disord 2022; 37(9): 1883–1892.

11.

Delva

Van Laere

Vandenberghe

Longitudinal imaging of regional brain volumes, SV2A, and glucose metabolism in Huntington’s disease. Mov Disord 2023; 38(8): 1515–1526.

12.

Mecca

O’Dell

Sharp

, et al. Synaptic density and cognitive performance in Alzheimer’s disease: a PET imaging study with [¹¹C]UCB-J. Alzheimers Dement 2022; 18: 2527–2536.

13.

Holmes

Honhar

Tinaz

, et al. Synaptic loss and its association with symptom severity in Parkinson’s disease. NPJ Parkinsons Dis 2024; 10: 42.

14.

Delva

Michiels

Koole

, et al. Synaptic damage and its clinical correlates in people with early Huntington disease: a PET study. Neurology 2022; 98: e83–e94.

15.

Naganawa

Nabulsi

, et al. First-in-human evaluation of ¹⁸F-SynVesT-1, a novel radioligand for PET imaging of synaptic vesicle protein 2A. J Nucl Med 2021; 62(4): 561–567.

16.

Naganawa

Gallezot

, et al. Noninvasive quantification of [¹⁸F]SynVesT-1 binding using simplified reference tissue model 2. Eur J Nucl Med Mol Imaging 2024; 52: 113–121.

17.

Michiels

Mertens

Thijs

, et al. Changes in synaptic density in the subacute phase after ischemic stroke: a ¹¹C-UCB-J PET/MR study. J Cereb Blood Flow Metab 2021; 42(2): 303–314.

18.

Koole

Van Aalst

Devrome

, et al. Quantifying SV2A density and drug occupancy in the human brain using [¹¹C]UCB-J PET imaging and subcortical white matter as reference tissue. Eur J Nucl Med Mol Imaging 2019; 46: 396–406.

19.

Finnema

Nabulsi

Mercier

, et al. Kinetic evaluation and test–retest reproducibility of [¹¹C]UCB-J, a novel radioligand for positron emission tomography imaging of synaptic vesicle glycoprotein 2A in humans. J Cereb Blood Flow Metab 2018; 38(11): 2041–2052.

20.

Mertens

Michiels

Vanderlinden

, et al. Impact of meningeal uptake and partial volume correction techniques on [¹⁸F]MK-6240 binding in aMCI patients and healthy controls. J Cereb Blood Flow Metab 2022; 41(11): 1–11.

21.

Giorgio

Soleimani-Meigooni

Janabi

, et al. Imaging synaptic density in aging and Alzheimer disease with [¹⁸F]SynVesT-1. J Nucl Med 2025; 66: 620–625.

22.

Lin

Bao

, et al. Head-to-head comparison of [¹⁸F]FDG, [¹⁸F]FMZ, and [¹⁸F]SynVesT-1 positron emission tomography imaging in patients with drug-resistant epilepsy. Eur J Nucl Med Mol Imaging 2025; 52: 2258–2266.

23.

Rossano

Toyonaga

Finnema

, et al. Assessment of a white matter reference region for ¹¹C-UCB-J PET quantification. J Cereb Blood Flow Metab 2020; 40(9): 1890–1901.

24.

Young

O’Dell

Naganawa

, et al. Validation of a simplified tissue-to-reference ratio measurement using SUVR to assess synaptic density alterations in Alzheimer disease with [¹¹C]UCB-J PET. J Nucl Med 2024; 65: 1782–1785.

25.

Muer

Didier

Wannebo

, et al. Sex differences in gray matter, white matter, and regional brain perfusion in young, healthy adults. Am J Physiol Heart Circ Physiol 2024; 327: H847–H858.

26.

Cai

Zhang

, et al. Synthesis and in vivo evaluation of [¹⁸F]UCB-J for PET imaging of synaptic vesicle glycoprotein 2A (SV2A). Eur J Nucl Med Mol Imaging 2019; 46(9): 1965.

27.

Badawi

Cherry

, et al. Performance characteristics of the NeuroEXPLORER, a next-generation human brain PET/CT imager. J Nucl Med 2024; 65(8): 1320–1326.

28.

Volpi

Toyonaga

Khattar

, et al. Exceptional brain PET images from the NeuroEXPLORER: scans with targeted radiopharmaceuticals and comparison to HRRT. Eur J Nucl Med Mol Imaging 2026; 53(3): 2110–2121.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

1.87 MB

0.00 MB

Quantification of 18 F-SynVesT-1 binding in healthy ageing and head-to-head comparison with 11 C-UCB-J

Abstract

Keywords

Introduction

Methods

Participants

PET data acquisition and processing

Statistical analysis

Results

Demographics

18F-SynVesT-1 kinetic modelling

Time stability of 18F-SynVesT-1 SUVR

Comparison of simplified SUVR measures using various reference regions

18F-SynVesT-1 binding in relation to age and sex

18F-SynVesT-1 head-to-head comparison to 11C-UCB-J

Discussion

Supplemental Material

sj-docx-1-jcb-10.1177_0271678X261454164 – Supplemental material for Quantification of 18F-SynVesT-1 binding in healthy ageing and head-to-head comparison with 11C-UCB-J

Footnotes

Acknowledgements

Author contributions

Funding

Declaration of conflicting interests

Ethical considerations

Consent to participate

ORCID iD

Data availability statement

Supplemental material

References

Supplementary Material

¹⁸F-SynVesT-1 kinetic modelling

Time stability of ¹⁸F-SynVesT-1 SUVR

¹⁸F-SynVesT-1 binding in relation to age and sex

¹⁸F-SynVesT-1 head-to-head comparison to ¹¹C-UCB-J