SV2A levels decreased significantly in both CSF and serum with AD progression

In this study, CSF SV2A levels in 14 patients with aMCI, 46 patients with AD, and 35 age-matched controls were first examined by the Simoa method. The results showed that SV2A levels in the CSF gradually decreased with the severity of dementia. Compared with the control group, mean levels of CSF SV2A in the aMCI and AD groups were significantly reduced by approximately 26.91% (p = 0.0013) and 56.96% (p < 0.0001), respectively. Compared with the mean level in the aMCI group, the mean level of CSF SV2A in the AD group was reduced by approximately 41.11% (p < 0.0001) (Fig. 2a). To reveal the relationship between CSF SV2A and cognitive ability in the AD group, a correlation analysis was also performed between CSF SV2A and cognitive scores, which showed a significant positive correlation between CSF SV2A levels and MMSE (r = 0.3928, p = 0.0002) and MOCA (r = 0.3905, p = 0.0002) (Fig. 2b-c).

Fig. 2

SV2A levels in CSF and serum at different stages of AD and the early diagnostic and differential diagnostic efficacy of them for AD. a. CSF SV2A levels at different stages of AD (Con = 35, aMCI = 14, AD = 46). b. Correlation of CSF SV2A with MMSE scores. c. Correlation of CSF SV2A with MOCA scores. d. Serum SV2A at different stages of AD (Con = 102, aMCI = 91, AD = 164). (e) Correlation of serum SV2A level with MMSE scores. (f) Correlation of serum SV2A level with MOCA scores. (g) Correlation analysis of CSF SV2A and serum SV2A (aMCI, n = 37; AD, n = 55). h–l. Diagnostic efficacy of CSF SV2A for Con vs. aMCI, Con vs. AD, aMCI vs. AD, VaD vs. AD, and PDD vs. AD. m–q. Diagnostic efficacy of serum SV2A for Con vs. aMCI, Con vs. AD, aMCI vs. AD, VaD vs. AD, and PDD vs. AD. CSF SV2A and serum SV2A were presented as means ± SEM. The significance of the between-group differences was determined using the Mann–Whitney U test. Partial correlation analyses were performed to assess the correlations between biomarkers and cognitive scores by controlling for confounders age and sex. *p < 0.05, **p < 0.01, ***p ≤ 0.001, ****p ≤ 0.0001. Abbreviations: 95% CI, 95% confidence interval; AD, Alzheimer’s disease; AUC, area under the curve; Con, control; aMCI, amnestic mild cognitive impairment; CSF, cerebrospinal fluid; PDD, Parkinson’s disease dementia; SV2A, synaptic vesicle glycoprotein 2 A; vs., versus

To further explore SV2A alterations in the blood, serum samples from 91 patients with aMCI, 164 patients with AD, and 102 age-matched controls were subsequently tested for SV2A using the Simoa method. Consistent with the trend in CSF, serum SV2A levels also progressively decreased with the progression of dementia, as evidenced by a significant reduction in the mean serum SV2A levels of approximately 55.32% (p < 0.0001) and 74.94% (p < 0.0001) in aMCI and AD, respectively, relative to controls, and a significant reduction of approximately 43.92% (p < 0.0001) in AD relative to aMCI (Fig. 2d). In the correlation analyses with cognitive scores, serum SV2A, which was similar to CSF SV2A, significantly and positively correlated with both MMSE (r = 0.2348, p < 0.0001) and MOCA (r = 0.2337, p < 0.0001) (Fig. 2e-f).

Based on the above results, SV2A showed high concordance between blood and CSF; therefore, a correlation analysis of 92 patients (including 37 patients with aMCI and 55 with AD) was performed in which both CSF and serum were tested. The results showed a significant positive correlation between serum SV2A and CSF SV2A (r = 0.5720, p < 0.0001) (Fig. 2g). The detailed data were presented in Table 1 and all these data were adjusted for age and sex.

CSF and serum SV2A had significant early diagnostic and differential diagnostic efficacy for AD

Since SV2A was closely associated with cognitive impairment in patients with AD, the early diagnostic and differential diagnostic ability of SV2A for AD was also analyzed. Diagnostic accuracy was evaluated using ROC curve analysis, and the Youden index was calculated to determine the best cutoff regarding sensitivity and specificity. Initially, CSF SV2A demonstrated significant diagnostic ability for aMCI (AUC = 78.8%, 95% CI = 0.647–0.891, sensitivity = 100.00%) (Fig. 2h) and differential diagnosis of aMCI from AD (AUC = 86.8%, 95% CI = 0.756–0.942) (Fig. 2j). CSF SV2A also showed high diagnostic performance for AD dementia (AUC = 93.5%, 95% CI = 0.857–0.978, sensitivity = 84.78%, specificity = 94.26%) (Fig. 2i). Although the mean level of CSF SV2A was lower in the VaD and PDD groups than in the cognitively unimpaired control group, the difference was not statistically significant (Table 1). Then, the CSF SV2A levels in the AD, VaD, and PDD groups were compared, revealing that the average CSF SV2A level was significantly lower in the AD group than in the other two groups, which demonstrated high diagnostic efficacy in the differential diagnosis of AD from VaD (AUC = 90.5%, 95% CI = 0.800–0.966, sensitivity = 84.78%, specificity = 92.31%) (Fig. 2k) and PDD (AUC = 91.8%, 95% CI = 0.817–0.974, sensitivity = 84.78%, specificity = 100.00%) (Fig. 2l).

Considering the invasive nature of CSF sampling, we next explored whether serum SV2A could be used as a screening indicator for early-stage AD. Surprisingly, like CSF SV2A, serum SV2A also demonstrated statistical significance in the diagnosis of aMCI (AUC = 74.1%, 95% CI = 0.674–0.802, sensitivity = 97.80%) (Fig. 2m) and differential diagnosis of aMCI from AD (AUC = 70.2%, 95% CI = 0.642–0.758) (Fig. 2o). Serum SV2A also demonstrated the same diagnostic efficacy as CSF SV2A in the diagnosis of AD (AUC = 86.6%, 95% CI = 0.820–0.905) (Fig. 2n). Compared with the significant decrease in AD, although the mean level of serum SV2A decreased in VaD and PDD relative to cognitively normal controls, it was not statistically significant. However, the mean level of serum SV2A, consistent with the trend in the CSF, was significantly lower in the AD group than in the VaD and PDD groups (Table 1), with good diagnostic efficacy in identifying AD from VaD (AUC = 82.3%, 95% CI = 0.764–0.872) (Fig. 2p) and PDD (AUC = 84.8%, 95% CI = 0.790–0.895) (Fig. 2q). All data are shown in Tables 1 and 2.

Table 2 Efficacy of SV2A for the early diagnosis and differential diagnosis of ADSV2A demonstrated high positivity rates in patients with aMCI who were negative for other biomarkers

The high diagnostic sensitivity (97.80%) in aMCI suggested that serum SV2A was valuable for the early screening of aMCI. To further support this point, we simultaneously tested other AD core biomarkers in the sera of the above diagnostic groups, including NfL, GFAP, and p-tau217 (Table 1). Then, we speculated whether serum SV2A could correct aMCI cases that were negative for other biomarkers. We counted the number of other biomarker-negative patients in the aMCI group and of which the number of SV2A-positive patients, respectively. Briefly, patients with serum NfL concentrations below its cutoff value (≤ 9.68 pg/mL in the aMCI group) (Table 3) were considered as serum NfL test-negative aMCI, whereas patients with serum SV2A concentrations below its cutoff value (≤ 5050.24 pg/mL in the aMCI group) (Table 2) were regarded as serum SV2A-positive aMCI. Statistical results showed that serum SV2A demonstrated an extremely high positivity rate of 100.00% in the NfL-negative cases of aMCI. By the same method, patients with concentrations below the cutoff value (≤ 7.67 pg/mL for GFAP and ≤ 3.24 pg/mL for p-tau217) (Table 3) were regarded as GFAP- and p-tau217 test-negative aMCI, respectively, and we found that serum SV2A was positive in 92.86% and 97.06% of GFAP and p-tau217-negative aMCI cases, respectively. In addition, we also calculated the positivity rates of serum SV2A in AD cases that were negative for other biomarkers by the same method, however, the rate was lower than that in the aMCI cases. All data on the positivity rate was presented in Table 4.

Table 3 Efficiency of serum SV2A in combination with other biomarkers in the diagnosis of aMCI or AD, and the differential diagnosis of AD from other dementiasTable 4 Positive rates of serum SV2A in cases negative for other biomarkersSerum SV2A combined with other biomarkers significantly improved the early diagnosis efficiency of AD

The above results indicated that SV2A demonstrated perfect complementarity with other biomarkers in the early diagnosis of AD, we further explored whether combining serum SV2A with other biomarkers could improve the diagnosis efficacy for aMCI. First, the correlation analysis showed that serum SV2A significantly negatively correlated with serum GFAP (r = − 0.1544, p = 0.0013) and serum p-tau217 (r = − 0.1355, p = 0.0049), respectively (Fig. 3a-b). Although serum SV2A did not demonstrate a significant correlation with NfL (Fig. 3c), the diagnostic models combining serum SV2A with NfL (GFAP or p-tau217) had significantly higher diagnostic AUCs than their corresponding single-indicator models (p < 0.01) (Fig. 3d-g). When serum SV2A was combined with the above three biomarkers, the AUC for aMCI was further improved to 91.8%, which was significantly higher than that of NfL, GFAP, or p-tau217 alone, as well as the diagnostic model combining the three biomarkers (p < 0.01) (Additional file 1: Table S1). The high sensitivity (97.80%) suggests that serum SV2A could be an excellent early screening biomarker for aMCI, whereas the low specificity (44.12%) will somewhat reduce its efficacy. Therefore, we tried to find indicators that could make up for this shortcoming. Compared with other indicators tested, serum p-tau217 had higher specificity (89.22%) for aMCI, and the diagnostic model combining serum SV2A and p-tau217 significantly increased the diagnostic specificity of aMCI to 95.1% by the serial test, which was further improved to 98% when serum SV2A was combined with the other three indicators (Table 3).

Fig. 3

Correlation analysis of serum SV2A with other biomarkers, and the diagnostic efficiency of serum SV2A combined with these biomarkers in aMCI and AD. Relationship of serum SV2A with (a) serum GFAP, (b) serum p-tau217, and (c) serum NfL. d. Diagnostic efficiency of serum NfL, GFAP, and p-tau217 in Con vs. aMCI. e. Diagnostic efficiency of serum NfL, GFAP, and p-tau217 in Con vs. AD. f. Diagnostic efficiency of serum NfL, GFAP, and p-tau217 in aMCI vs. AD. g. Diagnostic efficiency of serum SV2A + NfL, SV2A + GFAP, SV2A + p-tau217, and SV2A + NfL + GFAP + p-tau217 in Con vs. aMCI. h. Diagnostic efficiency of serum SV2A + NfL, SV2A + GFAP, SV2A + p-tau217, and SV2A + NfL + GFAP + p-tau217 in Con vs. AD. i. Diagnostic efficiency of serum SV2A + NfL, SV2A + GFAP, SV2A + p-tau217, and SV2A + NfL + GFAP + p-tau217 in aMCI vs. AD. Correlation analysis was performed using Spearman’s rank correlation coefficient. Abbreviations: AD, Alzheimer’s disease; aMCI, amnestic mild cognitive impairment; AUC, area under the curve; Con, healthy control; GFAP, glial fibrillary acidic protein; NfL, neurofilament light; p-tau217, phosphorylated tau; SV2A, synaptic vesicle glycoprotein 2 A; vs., versus

In addition to enhancing the diagnostic efficacy of aMCI in the cognitively normal populations, the combinations of serum SV2A with other biomarkers significantly improved the ability to identify aMCI from AD. Initially, among biomarkers tested, serum SV2A demonstrated the highest diagnostic AUC for identifying aMCI and AD, which was significantly higher than that of NfL (p = 0.0159) and p-tau217 (p = 0.0096) (Additional file 1: Table S2). After combining with serum SV2A, the AUCs of NfL, GFAP, and p-tau217 for the differential diagnosis between aMCI and AD significantly improved to 73.4%, 76.4%, and 72.7%, respectively (p < 0.01) (Fig. 3i). Although the combination of serum SV2A with other biomarkers could improve the sensitivity of distinguishing aMCI from AD to some extent, as a differential diagnostic marker, its specificity was more deserving of attention. By the serial test, the specificity of SV2A in identifying aMCI from AD was significantly improved when combined with other markers, especially with p-tau217, which increased the specificity from 80.22 to 100.0% (Table 3).

In the diagnosis of AD, the AUCs of the diagnostic models combining serum NfL, GFAP, and p-tau217 with serum SV2A were improved to 88.3%, 93.7%, and 94.2%, respectively, which were significantly higher than the AUCs of these biomarkers when used alone (p < 0.01). Excitingly, the combination of these four biomarkers resulted in a significant increase in the diagnostic AUC for AD to 96.2% (p < 0.01) (Fig. 3h and Additional file 1: Table S3). The specificity of 90.2% made serum SV2A an excellent diagnostic indicator for AD, and by the parallel test, its sensitivity significantly increased from 66.46 to 89.0%, 93.0%, and 93.0% after combining with NfL, GFAP, and p-tau217, respectively. The diagnostic model combining the above four biomarkers achieved a sensitivity of 100.0% by the parallel test (Table 3).

Serum SV2A combined with other biomarkers significantly improved the differential diagnosis efficiency of AD from non-AD dementia

As mentioned previously, serum SV2A demonstrated significant efficacy in the differential diagnosis of AD from other dementias. Subsequently, we examined the levels of serum NfL, GFAP, and p-tau217 in other dementias, further exploring the effect of combining multiple markers on the differential diagnosis of AD from other dementias. In the differential diagnosis of AD versus VaD, serum SV2A demonstrated a high diagnostic AUC, which was significantly higher than that of GFAP (p = 0.0087), NfL (p = 0.0086), and p-tau217 (p = 0.0010) (Additional file 1: Table S4). Serum NfL levels were significantly lower in AD than in VaD (Table 1), and the AUC for the differential diagnosis was 64.90% (Fig. 4a). Meanwhile, when combined with serum SV2A, the efficacy significantly increased to 85.6% (p = 0.0001) (Fig. 4d and Additional file 1: Table S4). Contrary to NfL levels, serum GFAP and p-tau217 levels were significantly higher in AD than in VaD (Table 1). When combined with serum SV2A, the AUC of GFAP for identifying AD from VaD significantly increased from 66.3 to 83.6% (p = 0.0007) (Fig. 4b-e), and that of p-tau217 from 63.1 to 81.9% (p = 0.0008) (Fig. 4c-f). The AUC of the diagnostic model combining the above four biomarkers in the differential diagnosis of AD from VaD significantly increased to 86.4% (Fig. 4g and Additional file 1: Table S4). We then combined serum SV2A with NfL, GFAP, or p-tau217 by the serial test and found that the specificity of identifying AD from VaD improved from 83.72 to 97.7%, 93.0%, or 93.0%, respectively (Table 3). On the other hand, the sensitivity values of the above diagnostic models improved to 80.5%, 86.6%, and 93.9% in the parallel test, respectively. The diagnostic model combining the above four biomarkers achieved a sensitivity of 98.2% by the parallel test and a specificity of 100.0% by the serial test (Table 3).

Fig. 4

Differential diagnostic efficiency of serum SV2A combined with other biomarkers in AD vs. VaD and AD vs. PDD. a–c. Differential diagnostic efficiency of serum NfL, GFAP, and p-tau217 in AD vs. VaD. d–g. Differential diagnostic efficiency of serum SV2A + NfL, SV2A + GFAP, SV2A + p-tau217, and SV2A + NfL + GFAP + p-tau217 in AD vs. VaD. h–j. Differential diagnostic efficiency of serum NfL, GFAP, and p-tau217 in AD vs. PDD. k–n. Differential diagnostic efficiency of serum SV2A + NfL, SV2A + GFAP, SV2A + p-tau217, and SV2A + NfL + GFAP + p-tau217 in AD vs. PDD. Abbreviations: 95% CI, 95% confidence interval; AD, Alzheimer’s disease; AUC, area under the curve; PDD, Parkinson’s disease dementia; SV2A, synaptic vesicle glycoprotein 2 A; VaD, vascular dementia; vs., versus

In the differential diagnosis of AD versus PDD, serum SV2A demonstrated a high diagnostic AUC, which was significantly higher than those of GFAP (p = 0.0430), NfL (p = 0.0009), and p-tau217 (p < 0.0001) (Additional file 1: Table S5). Although p-tau217 had no significant diagnostic value, the diagnostic AUC significantly increased to 85.5% when combined with serum SV2A (Fig. 4m). Serum NfL and GFAP levels were significantly higher in AD than in PDD (Table 1), and the AUCs in the differential diagnosis of AD from PDD significantly improved to 86.8% (p < 0.0001) (Fig. 4k) and 88.1% (p = 0.0001) (Fig. 4l), respectively, when combined with serum SV2A. The diagnostic model combining the above four biomarkers further improved the AUC for the differential diagnosis of AD and PDD to 89.6% (Fig. 4n), which was significantly higher than the value for the corresponding single biomarker (p < 0.01) (Additional file 1: Table S5). The specificity of 93.33% suggested that serum SV2A was an excellent biomarker of discriminating AD from PDD (Table 2), to improve its sensitivity, we combined serum SV2A with NfL, GFAP, or p-tau217 by the parallel test and found that the sensitivity of 64.63% increased to 83.5%, 93.3%, and 88.4%, respectively, while the sensitivity of the diagnostic model combining the four markers increased to 97.6% (Table 3).

Serum SV2A can effectively differentiate those at high risk of AD in the cognitively unimpaired population

Serum SV2A demonstrated significant diagnostic efficacy in the early diagnosis of AD, which prompted us to further explore the ability of the biomarker to identify those at high risk of AD in cognitively unimpaired individuals. We examined serum SV2A levels in 55 cognitively unimpaired APOE ε4 carriers and 60 cognitively unimpaired APOE ε4 non-carriers by the Simoa method, which showed that serum SV2A levels were significantly lower in APOE ε4 carriers than in APOE ε4 non-carriers (p < 0.0001), and the difference remained statistically significant even after correcting for age and sex (p = 0.003). In addition, we also examined the levels of serum NfL, GFAP, and p-tau217 in the above two groups. Although the mean levels of all three biomarkers were higher in APOE ε4 carriers than in APOE ε4 non-carriers, only the increase in GFAP was statistically significant (p = 0.024); unfortunately, the statistical significance disappeared after correcting for age and sex (p = 0.052). Detailed data were shown in Table 5.

Table 5 Clinical and demographic features of cognitively unimpaired subjects who have undergone APOE testing

Diagnostic accuracy was evaluated using ROC curve analysis, which showed that serum SV2A could significantly identify APOE ε4 carriers from APOE ε4 non-carriers, with a diagnostic AUC of 69.0% (95% CI = 0.597–0.773) (Fig. 5a), which was higher than that of other three biomarkers, where the difference with NfL was significant (p = 0.0193) and the difference with p-tau217 was approaching significance (p = 0.0688) (Additional file 1: Table S6). Although NfL, p-tau217, and GFAP had poor efficacy in distinguishing cognitively unimpaired APOE ε4 carriers from cognitively unimpaired APOE ε4 non-carriers, the diagnostic AUCs of the three biomarkers significantly improved to 67.1%, 72.8%, and 68.5%, respectively, when combined with serum SV2A, and the differential diagnostic AUC of the diagnostic model combining the four biomarkers increased to 74.5%, which was significantly higher than those of the corresponding single indicator, as well as the diagnostic model combining NfL, p-tau217 and GFAP (p < 0.05) (Fig. 5a-b and Additional file 1: Table S6). The sensitivity of 81.82% indicated that serum SV2A was suitable for screening individuals at risk for AD. To improve the specificity of SV2A, we combined it with GFAP, which demonstrated high specificity (90.0%). The specificity of this diagnostic model increased to 91.7% by the serial test, and the specificity further increased to 98.3% when SV2A was combined with the other three biomarkers. On the other hand, the results of the parallel test showed that serum SV2A combined with GFAP and p-tau217 increased the screening sensitivity for individuals at high risk of AD to 89.1% and 92.7%, respectively, and the sensitivity of the diagnosis with the above four biomarkers combined reached 100.0% (Table 6).

Fig. 5

Effectiveness of serum SV2A and other biomarkers for AD risk individual identification. (a) Efficacy of serum SV2A, NfL, GFAP, and p-tau217 in differentiating cognitively unimpaired APOE ε4 carriers from cognitively unimpaired APOE ε4 non-carriers, respectively. (b) Efficacy of serum SV2A + NfL, SV2A + GFAP, SV2A + p-tau217, and SV2A + NfL + GFAP + p-tau217 in differentiating cognitively unimpaired APOE ε4 carriers from cognitively unimpaired APOE ε4 non-carriers, respectively. Abbreviations: 95% CI, 95% confidence interval; APOE ε4 −/−, APOE ε4 non-carriers; APOE ε4 +/−, APOE ε4 carriers; AUC, area under the curve; GFAP, glial fibrillary acidic protein; NfL, neurofilament light; p-tau217, phosphorylated tau; SV2A, synaptic vesicle glycoprotein 2 A; vs., versus

Table 6 The efficiency of serum SV2A combined with other biomarkers in discriminating cognitively unimpaired APOE ε4 carriers from APOE ε4 non-carriers

The Youden index was calculated to determine the best cutoff value between APOE ε4 carriers and APOE ε4 non-carriers for each biomarker. Patients with SV2A concentrations below its cutoff value (≤ 4413.00 pg/mL) were defined as serum SV2A-positive cases. Similarly, patients with concentrations above the cutoff values (> 1.1 pg/mL for NfL, > 16.6 pg/mL for GFAP, and > 1.8 pg/mL for p-tau217) were defined as NfL, GFAP, and p-tau217–positive cases, respectively (Table 6). Then, we counted the number of SV2A-, GFAP-, NfL-, and p-tau217-positive cases in the AD high-risk group (APOE ε4 carriers) and found that a total of 45 cases were positive for SV2A, with a positivity rate of 81.82%, which was higher than those of GFAP (32.73%) and p-tau217 (36.36%) (Table 7). Although the positivity rate of NfL was 96.36%, the indicator with an extremely low specificity (18.33%) was not a statistically significant early warning marker of AD (Table 6). Then, we counted the number of SV2A-positive cases in GFAP-, NfL-, and p-tau217-negative cases among APOE ε4 carriers (≤ 1.1 pg/mL for NfL, ≤ 16.6 pg/mL for GFAP, and ≤ 1.8 pg/mL for p-tau217), and the results showed that the positivity rate of SV2A in GFAP-, NfL-, and p-tau217-negative cases reached 83.78%, 100.00%, and 88.57%, respectively. Finally, the positivity rate significantly increased to 89.09%, 92.73%, and 100.00% among APOE ε4 carriers when serum SV2A-positive cases were counted together with positive cases of GFAP, p-tau217 and NfL, respectively (Table 7).

Table 7 Positive rates of SV2A in APOE ε4 carriers who were negative for other biomarkersSV2A affect Aβ pathology in APP/PS1 mice

Aβ pathology was recognized as the key factor in the development of AD. To assess the value of SV2A in predicting Aβ pathology, SV2A overexpressing APP/PS1 mice and control APP/PS1 mice were constructed by brain stereotactic injection technique, respectively. Firstly, immunofluorescence staining with 6E10 antibody, which recognized the first 16 amino acids of the Aβ sequence, was conducted to identify Aβ plaques in brain sections of above two groups (Fig. 6a-b). The results indicated that the number and size of Aβ plaques and 6E10 antibody staining intensity in the brain tissues of SV2A overexpressing APP/PS1 mice was significantly reduced compared with control mice (Fig. 6c). Meanwhile, levels of metabolites of Aβ precursor protein (APP) in the serum of the above two groups of mice were measured by ELISA, and the results showed that the serum levels of both Aβ40 (Fig. 6d) and Aβ42 (Fig. 6e) were significantly reduced in SV2A overexpressing APP/PS1 mice compared to control mice. Therefore, the above results revealed that SV2A could significantly inhibit Aβ pathology in APP/PS1 mice.

Fig. 6

Effect of SV2A on Aβ pathology in APP/PS1 mice. (a) Fluorescence intensity of 6E10 in hippocampal regions of SV2A overexpressing APP/PS1 mice and control APP/PS1 mice. (b) Fluorescence intensity of 6E10 in cortical regions of SV2A overexpressing APP/PS1 mice and control APP/PS1 mice. (c) Quantification of amyloid plaques per mm2 area, amyloid plaque size (µm2), and the ratio of fluorescence intensity of 6E10 to DAPI in brain sections. AAV-SV2Aoe APP/PS1 mice group, n = 3; AAV-Con APP/PS1 mice group, n = 3; 3 fields of view per group. (d) The serum levels of Aβ40 in SV2A overexpressing APP/PS1 mice and control mice. (e) The serum levels of Aβ42 in SV2A overexpressing APP/PS1 mice and control mice. Scale bar: 50 μm. Data were presented as mean ± SD. All dot plots: t-test. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001