Dynamic changes of CSF clusterin levels across the Alzheimer’s disease continuum

This study was based on publicly available data from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) database. ADNI aims to examine whether clinical, imaging and genetic assessments as well as CSF biomarkers can be integrated to detect and track AD as soon as possible. Any neurological disease patients were excluded from the database, except suspected AD. The ages of participants are in 55 to 90. On the ADNI website, more detailed information about inclusive or exclusive criteria can be found. In accordance with the Declaration of Helsinki, written informed consent was acquired from all the participants or their representatives. ADNI was approved by the institutional review boards of all the participating institutions. For latest information, see ADNI website. [22, 23].

We included 285 participants with baseline data on CSF clusterin, CSF amyloid-β (1–42) (Aβ₄₂), phosphorylated-Tau (p-tau), and total-tau (t-tau) levels from the ADNI database. Nine participants with values outside the mean ± 3 standard deviations (SD) were excluded, leaving 276 participants for further analysis.

CSF Aβ₄₂, p-tau, t-tau were all analyzed using the INNO-BIA Alz-Bio3 immunoassay. These within-batch precision values were all below 10% (respectively Aβ₄₂: 5.1 to 7.8%, p-tau: 5.1 to 8.8%, t-tau: 4 to 9.8%). At the same time, these CSF core biomarkers were analyzed by the fully automated Elecsys immune assay platform® at the University of Pennsylvania. More detailed information about Elecsys method measuring AD biomarkers in ADNI is provided elsewhere [24]. All CSF biomarkers assays were repeated and averaged. The quantification of CSF clusterin was conducted using liquid chromatography-tandem mass spectrometry with multiple reaction monitoring approach (LC/MS-MRM) [25]. The whole panel consist of 567 petides representing 221 proteins, and the peptide sequence used for clusterin was IDSLLENDR in our research. We obtained data from ADNI at http://​adni.​loni.​usc.​edu/​.

Following 2018 NIA-AA criteria, each ADNI participant was assigned into groups by the ATN framework [21]. At present, this classification system serves research purposes rather than clinical purposes. Additionally, ATN classification system is useful for observational studies since it stresses the predictive value of biomarkers [26]. The primary criteria for AD classification are dependent on underlying pathophysiological damage rather than clinical manifestations. In the ATN classification system, “A” refers to aggregated Aβ (CSF Aβ₄₂ or amyloid PET); “T” refers to aggregated tau (CSF p-tau or tau PET); and “N” refers to neurodegeneration (CSF t-tau, FDG-PET or MRI) [21, 27]. In our study, we implemented this classification based on CSF Aβ₄₂ levels (A); CSF p-tau levels (T); CSF t-tau (N). To reduce comparison groups numbers, the tau pathology group (T) and neurodegeneration group (N) were merged, given that CSF p-tau and t-tau are closely correlated. ‘(TN)-’ profile was defined as both normal range of CSF p-tau and t-tau and ‘(TN)+’ profile was defined as p-tau or t-tau with abnormal range. We dichotomized biomarkers according to use previously published cutoffs as normal (negative) or abnormal (positive). CSF Aβ₄₂ < 976.6 pg/mL, CSF p-tau > 21.8 pg/mL, CSF t-tau > 245 pg/ml were respectively defined as A positive (A+), T positive (T+) and N positive (N+) [28]. Then, we obtained four different groups, containing A–(TN)–, A+(TN)–, A+(TN)+, A-(TN)+. Specifically, participants with normal pathology defined as “A–(TN)–”, abnormal amyloid with normal t-tau and p-tau defined as “A+(TN)–”, abnormal amyloid and abnormal levels of t-tau/p-tau defined as “A+(TN)+”, and suspected non-AD pathology defined as A-(TN)+. In addition, individuals were divided into the biomarker normal group (A–(TN)–) and AD continuum group (A+(TN)– and A+(TN)+).

The data downloaded from the ADNI database showed that the concentration of CSF clusterin displayed an approximate normal distribution, which was presented in the Q-Q plot (Supplementary Fig. 1). CSF Aβ₄₂, CSF p-tau and t-tau levels were log-transformed to obtain normal distributions. We excluded values outside the mean ± 3 standard deviations (SD) to eliminate the influence of extreme values. Totally, nine individuals were excluded in this step. To explore the differences in CSF biomarkers and sociodemographic data, we employed one-way analyses of variance (ANOVA) for continuous variables and chi square for dichotomous variables (such as APOE ε4 status and gender). Subsequently, to test the levels of CSF clusterin in different ATN groups, we employed an analysis of covariance (ANCOVA) followed by Bonferroni post hoc analyses. Thus, the threshold P values of groups differences for statistical significance was set at 0.008 (= 0.05/6). Finally, in the whole cohort, the biomarker normal group and the AD continuum group, a multiple linear regression was performed to study the associations of the concentration of CSF clusterin with CSF core pathological biomarkers, adjusting for age, diagnosis, gender, education, and APOE ε4 status. P <  0.05 was defined as statistical significance the correlation analysis. We used R software version 4.1.0 and IBM SPSS Statistics to perform all statistical analyses.

The demographics and clinical features of participants at baseline are summarized in the Table 1. Our current study, 276 participants were included from the publicly available ADNI database (A-(TN)-, n = 50; A+(TN)-, n = 39; A+(TN)+, n = 147; A-(TN)+, n = 40). The study population had a female population of 39.5%, an APOE ε4 positive percentage of 47.8%, an average age of 75.11 ± 6.86 years, and 15.72 ± 2.99 years of education. Four groups showed statistically significant differences in the distribution of APOE ε4 status and MMSE score (both P <  0.001) rather than age, education level, and gender. The proportion of MMSE scores and APOE ε4 are dependent on the ATN framework, of which the highest percentage of APOE ε4 and lowest MMSE scores are in A+ profiles. Compared with the other three groups, A+(TN)- group showed the lowest levels of CSF clusterin.

To assess the levels of CSF clusterin across the AD continuum, we employed ATN classification framework. Significant differences were shown by one-way ANCOVA after Bonferroni correction among the four groups (Fig. 1). As mentioned above, the A+(TN)- group had the lowest concentration of CSF clusterin. And compared to the A+(TN)- group, increased level of CSF clusterin was in the A+(TN) + group (P <  0.001 after Bonferroni correction). Considering age, gender, APOE ε4, education level all influence AD, we conducted another ANCOVA after adjusting for all these factors, which yielded the same results as previous studies.

We also repeated ANCOVA in groups classified by Aβ pathology and tau pathology status or in groups classified by Aβ pathology and neurodegeneration status. The results of the two analyses are presented in the Supplementary Fig. 3 and Supplementary Fig. 4, which are consistent with those based on ATN classification system. This consistency suggests that the preceding grouping is reasonable.

Our results suggest that during AD progression, CSF clusterin levels are different, which may be associated with the pathological changes in ATN biomarkers.

Participants who had abnormal tau pathology or neurodegeneration without amyloidosis(A-(TN) + group)were considered as suspected AD. The results of the Bonferroni post hoc test showed CSF clusterin levels were different among groups. As shown in Fig. 1, the higher level of CSF clusterin was in the A-(TN) + group compared to the A-(TN)- group (no pathology, P <  0.001 after Bonferroni correction), the A+(TN)- group (amyloid-only pathology, P <  0.001 after Bonferroni correction), and the A+(TN) + group (abnormal amyloid and abnormal t-tau/p-tau, P <  0.001 after Bonferroni correction) after adjusting for age, diagnosis, education, gender, and APOE ε4 status. This result suggests that the elevation of CSF clusterin level may be associated with tau pathology/neurodegeneration.

Finally, to study the associations between CSF clusterin and the core AD biomarkers based on ATN system, we employed the linear regression methods adjusting for age, diagnosis, education, gender, and APOE ε4 status. We excluded participants in the A-(TN) + group in this analysis. These results were presented in Supplementary Table 1. In all the participants (n = 237), CSF clusterin was positively associated with CSF Aβ₄₂ (β = 0.040, P <  0.001) (Fig. 2a), CSF p-tau (β = 0.325, P <  0.001) (Fig. 2d), and CSF t-tau (β = 0.346, P < 0.001) (Fig. 2g). We further investigated these correlations in the subgroups, including biomarker normal and AD continuum groups. Figure 2b showed no significant association was observed between CSF clusterin and CSF Aβ₄₂ in the biomarker normal group (β = 0.002, P = 0.763). In AD continuum groups, there was a positive relationship between CSF clusterin and CSF Aβ₄₂ in Fig. 2c. Results showed the CSF clusterin were positively associated with CSF t-tau and CSF p-tau both in biomarker normal group and AD continuum group (Fig. 2h, e, i, f). We studied the association of CSF clusterin with CSF Aβ₄₂ among A-TN- and A + TN- subjects. There was a positive relationship between CSF clusterin and CSF Aβ₄₂ (β = 0.026, P < 0.005). We repeated the analysis after excluding the outliers, which yielded the similar results Supplementary Table 2.