Characteristics of Participants in the CABLE Study
We included 651 non-demented elders from the CABLE study, consisting of 457 CN controls (238 females, 60.54 ± 10.46 years) and 194 MCI patients (109 females, 63.6 ± 9.72 years) (Table 1). The CN individuals were significantly younger and more educated, and had significantly lower levels of CSF p-tau and t-tau, compared to the MCI participants.
Table 1. Demographics of the study population in CABLE a
|
CN (n = 457)
|
MCI (n = 194)
|
P value
|
AGE, mean (SD), years
|
60.93 (10.55)
|
65.44 (10.01)
|
<0.001
|
Female, n (%)
|
269 (58.9)
|
109 (56.2)
|
0.59
|
APOE ε4 genotype carriers, n (%)
|
69 (15.1)
|
35 (18.0)
|
0.41
|
Education, mean (SD), years
|
10.38 (6.12)
|
8.56 (4.23)
|
<0.001
|
CSF α-synuclein, mean (SD), ng/l
|
1466.73 (813.99)
|
1501.19 (914.13)
|
0.61
|
CSF PTAU , mean (SD), ng/l
|
38.11 (9.69)
|
40.04 (12.42)
|
0.03
|
CSF TAU, mean (SD), ng/l
|
173.3 (77.96)
|
191.02 (122.57)
|
0.03
|
CSF ABETA42, mean (SD), ng/l
|
160.01 (91.51)
|
162.10 (105.53)
|
0.81
|
Aβ, β-amyloid; CN, cognitively normal; CSF, cerebrospinal fluid; MCI, mild cognitive impairment; MMSE, Mini-Mental State Examination; p-tau, phosphorylated tau; t-tau, total tau.
aP values from the Kruskal-Wallis test or Fisher exact test.
CSF α-synuclein and established AD biomarkers in the CABLE study
In the CABLE study, we examined the concentrations of CSF α-synuclein and other established AD biomarkers (CSF Aβ, p-tau and t-tau) and tested their relationships (Table 2). We found that the level of CSF α-synuclein was positively associated with the CSF t-tau (β = 0.56, P <0.001) and p-tau (β = 0.35, P <0.001) among the non-demented participants. However, there was no association between CSF α-synuclein and CSF Aβ level at baseline. In addition, the same associations were found in the CN group and the MCI group (Table 2).
Table 2. Correlations CSF α-synuclein and other biochemical markers in CABLE a
|
All participants
|
CN
|
MCI
|
β coefficient
|
P value
|
β coefficient
|
P value
|
β coefficient
|
P value
|
CSF t-tau
|
0.56
|
<0.001
|
0.38
|
<0.001
|
0.67
|
<0.001
|
CSF p-tau
|
0.35
|
<0.001
|
0.27
|
<0.001
|
0.40
|
<0.001
|
CSF Aβ42
|
-0.02
|
0.97
|
-0.01
|
0.82
|
-0.07
|
0.69
|
Aβ, β-amyloid; CABLE, Chinese Alzheimer's Biomarker and Lifestyle; CN, cognitively normal; CSF, cerebrospinal fluid; MCI, mild cognitive impairment; p-tau, phosphorylated tau; t-tau, total tau.
aData are β coefficients (with P values) from linear regression models for correlations between CSF α-synuclein and other biomarkers, adjusted for age, gender, educational level and APOE ε4 genotype. Models were tested in the whole cohort and in individual diagnostic groups.
Characteristics of participants in ADNI
Three hundred and eighty-two subjects from the ADNI database were included (Table 3). This cohort consisted of 109 CN controls (54 females, 75.63 ± 5.22 years), 117 sMCI patients (37 females, 74.34 ± 7.60 years), 66 pMCI patients (25 females, 74.21± 7.58 years) and 90 AD patients (39 females, 74.89 ± 7.72 years). According to the new “ATN” scheme, 258 A+ (220 A+T+) patients and 124 A- (96 A-T-) controls were included. As expected, the AD group had the highest frequency of the APOE ε4 allele (69.23%) and the CN controls group had the lowest frequency (23.85%). There was no significant difference in the educational level (P = 0.16) or age (P = 0.53) among these four groups. Furthermore, AD patients had lower MMSE scores compared with the MCI patients and CN controls (P <0.01).
Table 3. Demographics for the Study Population in ADNI
|
CN (n = 109)
|
sMCI (n = 117)
|
pMCI (n = 66)
|
AD (n = 90)
|
Age, mean (SD), years
|
75.63(5.22)
|
74.34(7.60)
|
74.21(7.58)
|
74.89(7.72)
|
Female, n (%)
|
54(49.54)
|
37(31.62)
|
25(36.76)
|
39(44.32)
|
APOE ε4 genotype carriers, n (%)
|
26(23.85)
|
55(47.00)
|
42(61.76)
|
63(69.23)
|
CSF α-synuclein, mean (SD), ng/L
|
0.46(0.17)
|
0.54(0.22)
|
0.56(0.20)
|
0.61(0.24)
|
MMSE score, mean (SD)
|
29.07(1.05)
|
27.15(1.64)
|
26.58(1.77)
|
23.39(1.80)
|
CSF Aβ42, mean (SD), ng/L
|
208.70(52.36)
|
174.69(55.28)
|
148.75(41.52)
|
143.99(38.31)
|
CSF t-tau, mean (SD), ng/L
|
69.08(29.85)
|
97.31(64.77)
|
112.00(41.52)
|
122.83(57.09)
|
CSF p-tau, mean (SD), ng/L
|
25.04(13.93)
|
32.76(18.31)
|
39.50(17.18)
|
41.48(19.73)
|
Hippocampus volume, mm3
|
6648.16(766.59)
|
5964.07(986.76)
|
5522.46(1044.15)
|
5217.39(1043.40)
|
Abbreviations: Aβ, β-amyloid; AD, Alzheimer disease dementia; CN, cognitively normal; CSF, cerebrospinal fluid; sMCI, stable mild cognitive impairment; pMCI, progressive mild cognitive impairment; MMSE, Mini-Mental State Examination; p-tau, phosphorylated tau; t-tau, total tau.
CSF α-synuclein and established AD biomarkers in ADNI
In the ADNI database, we found that the high CSF α-synuclein levels were associated with the high CSF t-tau (β = 0.27, P <0.001) and p-tau (β = 0.36, P <0.001) in the whole cohort. However, there was no association between CSF α-synuclein and CSF Aβ level at baseline. The same results were obtained in the MCI group (CSF t-tau: β = 0.29, P <0.001, CSF p-tau: β = 0.33, P <0.001) and CN controls (CSF t-tau: β = 0.2, P <0.001, CSF p-tau: β = 0.32, P <0.001).In addition, the CSF α-synuclein concentration was associated with CSF NFL concentration in non-demented elders (β = 0.12, P <0.001). However, there was no association between CSF α-synuclein and plasma NFL (Table 4, Fig. S1).
Table 4. Modelling the association of CSF biomarkers on AD biomarkers and clinical outcomes in ADNIa
|
All Participants
|
MCI
|
CN
|
Cross-sectional (MR)
|
β coefficient
|
P value
|
β coefficient
|
P value
|
β coefficient
|
P value
|
CSF t-tau
|
0.27
|
<0.001
|
0.29
|
<0.001
|
0.20
|
<0.001
|
CSF p-tau
|
0.36
|
<0.001
|
0.33
|
<0.001
|
0.32
|
<0.001
|
CSF Aβ42
|
-0.03
|
0.33
|
-0.04
|
0.32
|
0.006
|
0.86
|
CSF NFL
|
0.12
|
<0.001
|
0.11
|
0.04
|
0.03
|
0.45
|
Plasma NFL
|
0.04
|
0.27
|
0.02
|
0.73
|
-0.04
|
0.53
|
Longitudinal (MELM)
|
|
|
|
|
|
|
Hippocampus
|
-0.008
|
0.001
|
-0.007
|
0.04
|
-0.003
|
0.17
|
Ventricles
|
0.006
|
0.13
|
0.005
|
0.36
|
0.003
|
0.43
|
Cox (Hazard ratio)
|
Statistic
|
P value
|
|
MCI-to-AD dementia conversion
|
1.53(1.15-2.0)
|
0.004
|
Abbreviations: CN, cognitively normal; CSF, cerebrospinal fluid; MCI, mild cognitive impairment; p-tau, phosphorylated tau; t-tau, total tau; Cox, Cox proportional hazard model; MELM, mixed effects linear model; MR, multiple regression.
aAll models were adjusted for age, gender, educational level, APOE ε4 genotype and intracranial volume (for MRI only). Models were tested in the whole cohort and in individual diagnostic groups.
CSF α-synuclein in different diagnostic groups in ADNI
The level of CSF α-synuclein showed a trend of increase with the progression of disease stage. The CSF α-synuclein concentration was significantly higher in the AD and pMCI groups than in the CN controls (P <0.0001 and P <0.001respectively) and the sMCI group (P = 0.02 and P= 0.04, respectively) (Fig. 1a). In addition, the A+ AD group had higher CSF α-synuclein levels than the A- controls (P <0.001), A+ controls (P <0.001), and A- MCI group (P <0.001) (Fig. 1b). The A+ MCI had higher CSF α-synuclein levels than the A- controls (P <0.01), A+ controls (P <0.01), and the A- MCI group (P = 0.02). The CSF α-synuclein level was also significantly different between the A+T+ group and the A-T- group (P <0.0001) (Fig. 1c).
We generated receiver-operating curves based on the logistic regression models adjusted for age at baseline, gender, educational level and APOE ε4 genotype to assess the predictive value of CSF α-synuclein alone and in combination with other established AD biomarkers for the risk of conversion to AD. The area under the curve (AUC) of the baseline model containing CSF α-synuclein, age at baseline, gender, educational level and APOE ε4 genotype was 0.76 in predicting the onset of AD among the CN controls, and the AUC was further increased by the inclusion of CSF tau/Aβ ratio (AUC = 0.88) (Fig. S2). As expected, the baseline model showed a similar predicting value for the onset of pMCI among the CN controls (Fig. S3). In the A- group, this baseline model showed a good predictive value for the risk of conversion to A+ status (AUC = 0.77), and inclusion of CSF t-tau (AUC = 0.88) and p-tau (AUC = 0.92) further enhanced this predictive value (Fig. S4). Furthermore, the baseline model performed best when the participants were grouped by Aβ deposition and pathology (AUC = 0.84). We also detected that CSF α-synuclein added value for diagnosis prediction (Fig. S5).
CSF α-synuclein, longitudinal neuroimaging change and progression in ADNI
Next, the linear mixed-effects models were utilized to test the associations between baseline CSF α-synuclein concentration and subsequent disease progression, after adjustment for age, gender, educational level, diagnosis, and APOE ε4 genotype. The baseline CSF α-synuclein concentration was found to be significantly associated with the hippocampal volume (β = -0.008, P = 0.001 longitudinally) (Table 4, Fig. 2 (left)).
Fig. 2 (right) presents the results of a Kaplan-Meier analysis. The cox proportional hazards model was developed to estimate the predictive value of CSF α-synuclein for the conversion risk from MCI to incidence of AD dementia, after controlling for baseline age, gender and years of education. MCI individuals with high CSF α-synuclein levels would satisfy the diagnostic criteria for AD at a comparatively earlier interval (HR 2.79, 95% CI 1.14–6.9, P = 0.03) (Table 4).