Skip to main content
NCBI home page
Science Advances logoLink to Science Advances
. 2024 Apr 3;10(14):eadk3674. doi: 10.1126/sciadv.adk3674

CSF biomarkers of immune activation and Alzheimer’s disease for predicting cognitive impairment risk in the elderly

Francis Shue 1, Launia J White 2, Rachel Hendrix 3,†, Jason Ulrich 3, Rachel L Henson 3, William Knight 3, Yuka A Martens 1,‡, Ni Wang 1, Bhaskar Roy 1, Skylar C Starling 1, Yingxue Ren 2, Chengjie Xiong 4, Yan W Asmann 2, Jeremy A Syrjanen 5, Maria Vassilaki 5, Michelle M Mielke 5,§, Jigyasha Timsina 6, Yun Ju Sung 6, Carlos Cruchaga 6, David M Holtzman 3, Guojun Bu 1,, Ronald C Petersen 7, Michael G Heckman 2, Takahisa Kanekiyo 1,*
PMCID: PMC10990276  PMID: 38569027

Abstract

The immune system substantially influences age-related cognitive decline and Alzheimer’s disease (AD) progression, affected by genetic and environmental factors. In a Mayo Clinic Study of Aging cohort, we examined how risk factors like APOE genotype, age, and sex affect inflammatory molecules and AD biomarkers in cerebrospinal fluid (CSF). Among cognitively unimpaired individuals over 65 (N = 298), we measured 365 CSF inflammatory molecules, finding age, sex, and diabetes status predominantly influencing their levels. We observed age-related correlations with AD biomarkers such as total tau, phosphorylated tau-181, neurofilament light chain (NfL), and YKL40. APOE4 was associated with lower Aβ42 and higher SNAP25 in CSF. We explored baseline variables predicting cognitive decline risk, finding age, CSF Aβ42, NfL, and REG4 to be independently correlated. Subjects with older age, lower Aβ42, higher NfL, and higher REG4 at baseline had increased cognitive impairment risk during follow-up. This suggests that assessing CSF inflammatory molecules and AD biomarkers could predict cognitive impairment risk in the elderly.

Risk of cognitive impairment is associated with age and CSF levels of Aβ42, NfL, and REG4.

INTRODUCTION

Alzheimer’s disease (AD) is the most common form of dementia in older adults. Approximately 60 to 80% of dementia cases are diagnosed as AD, which is ranked as the sixth leading cause of death in the United States (1). The framework for AD classification is proposed to use fluid biomarker measurements and neuroimaging for amyloid-β (Aβ) deposition (A), pathologic tau (T), and neurodegeneration (N) (2). Cerebrospinal fluid (CSF) biomarkers associated with AD pathology include Aβ42, total tau, and phosphorylated tau181 (p-tau181) (3). In addition, neurofilament light chain (NfL) (4, 5) and several synaptic proteins, including neurogranin (Ng) (6, 7) and synaptosomal-associated protein 25 (SNAP25) (8) have been demonstrated to increase in CSF from patients with AD as sensitive biomarkers for axonal degeneration and synaptic damage, respectively.

Clinical AD diagnosis accuracy using medical history, neuropsychological testing, neuroimaging, and/or fluid biomarkers has improved substantially (9). However, a milieu of changes occurs in the brain before detectable AD symptoms. There is an urgent need to establish effective biomarkers to predict cognitive decline in the presymptomatic stage to identify patients and initiate preventive/therapeutic interventions in the early stages of disease. Of note, CSF immune activation markers have been recognized as potential biomarkers associated with the progression of cognitive impairment as the immune system substantially contributes to AD pathogenesis (10). CSF chitinase-3 like-1, cartilage glycoprotein-39 (YKL40) (11, 12) and soluble triggering receptor expressed on myeloid cells 2 (sTREM2) (13, 14) associate with astrocyte and microglia activation, respectively, and increase from the stage of preclinical AD to mild cognitive impairment (MCI).

To further explore the relationship between brain inflammation and age-related cognitive decline, we measured 365 CSF immune activation markers using proximity extension assay technology in 298 cognitively unimpaired participants in the Mayo Clinic Study of Aging (MCSA) (15). Established AD biomarkers (Aβ42, total tau, p-tau181, NfL, Ng, SNAP25, and YKL40) in CSF were also assessed. The aims of this study were to evaluate associations of CSF immune activation markers and established AD biomarkers with APOE genotype, clinical/epidemiological characteristics, and longitudinal progression of clinical status using the Clinical Dementia Rating (CDR) scale (16) conversion from cognitively unimpaired to impaired, as well as to understand relationships between the immune activation markers and established biomarkers.

RESULTS

Impact of APOE on AD biomarkers and immune activation markers in CSF among cognitively unimpaired participants

A total of 298 cognitively unimpaired individuals (CDR = 0) from the MCSA who had CSF samples available, were over 65 years of age, were non-Hispanic and white, were unrelated, and had APOE genotype available were included in this longitudinal cohort study. Information was collected regarding APOE genotype, age, sex, years of education, body mass index (BMI), smoking, hypertension, diabetes, and dyslipidemia. APOE genotype was ε2/ε3 in 40 subjects, ε3/ε3 in 184 subjects, ε3/ε4 in 70 subjects, and ε4/ε4 in 4 subjects. Because of the small number of ε4/ε4 subjects, APOE genotype was categorized as APOE2 (ε2/ε3, N = 40), APOE3 (ε3/ε3, N = 184), or APOE4 (ε3/ε4 or ε4/ε4, N = 74) in all analyses. The median age was 74 years (65 to 93 years), and 178 participants (59.7%) were male. There were no major differences in demographics between APOE groups at baseline (Table 1). A total of 372 CSF analytes were assessed, including seven AD biomarkers (Aβ42, total tau, p-tau181, NfL, SNAP25, Ng, and YKL40) and 365 immune activation markers. After correcting for multiple testing (P < 0.00013 considered as significant), CSF Aβ42 was significantly lower in APOE4 subjects compared to APOE3 subjects in multivariable analysis adjusting for participant characteristics (β: −279.43, P = 6.22 × 10−06), while CSF SNAP25 was higher in APOE4 subjects (β: 0.38, P = 1.05 × 10−09) (Table 2). Even after further adjusting models for all other AD biomarkers, APOE4 was still significantly associated with CSF Aβ42 (β: −316.43, P = 4.97 × 10−08) and SNAP25 (β: 0.39, P = 6.13 × 10−12). In addition, although not statistically significant after correcting for multiple testing, suggestive differences (P < 0.01) were observed between APOE2 and APOE3 subjects for FCRL3 (β: −0.53, P = 0.006) and between APOE4 and APOE3 subjects for p-tau181 (β: 4.14, P = 0.004), CSF3 (β: −0.57, P = 0.002), interleukin-22RA1 (IL-22RA1) (β: −0.30, P = 0.008), and IL-24 (β: −0.14, P = 0.01) (Table 2 and Fig. 1). When APOE was assessed as an ordinal variable (APOE2 = 1, APOE3 = 2, APOE4 = 3), significant associations with Aβ42 (β: −193.73, P = 5.12 × 10−06) and SNAP25 (β: 0.19, P = 5.12 × 10−06) remained consistent, with suggestive associations noted for HCLS1 (β: −0.12, P = 0.007) and PRSS8 (β: −0.15, P = 0.005) (table S1).

Table 1. Summary of participant characteristics overall and separately according to APOE group.

P values result from a Kruskal-Wallis rank sum test (continuous variables) or Fisher’s exact test (categorical variables). APOE2, ε2/ε3 (N = 40); APOE3, ε3/ε3 (N = 184); APOE4, ε3/ε4 (N = 70) and ε4/ε4 (N = 4).

Median (minimum, maximum) or no. (%) of subjects
All (N = 298) APOE2 (N = 40) APOE3 (N = 184) APOE4 (N = 74) P value
Age (years) 74.0 (65.0, 92.7) 74.1 (65.3, 90.4) 74.0 (65.0, 92.7) 73.7 (65.1, 85.9) 0.40
Sex (male) 178 (59.7%) 29 (72.5%) 109 (59.2%) 40 (54.1%) 0.16
Education (years) 14 (6, 20) 13.5 (6.0, 20.0) 14.0 (8.0, 20.0) 16.0 (8.0, 20.0) 0.081
BMI 28.1 (18.1, 46.6) 28.8 (19.0, 46.6) 27.9 (18.2, 42.8) 28.0 (18.1, 44.3) 0.39
BMI > 30 104 (34.9%) 15 (37.5%) 61 (33.2%) 28 (37.8%) 0.69
Smoking (former or current) 157 (52.7%) 20 (50.0%) 97 (52.7%) 40 (54.1%) 0.92
Hypertension 196 (65.8%) 31 (77.5%) 116 (63.0%) 49 (66.2%) 0.22
Diabetes 54 (18.1%) 10 (25.0%) 33 (17.9%) 11 (14.9%) 0.42
Dyslipidemia 247 (82.9%) 28 (70.0%) 155 (84.2%) 64 (86.5%) 0.079
Open in a new tab

Table 2. Comparisons of AD biomarkers and immune activation markers in CSF between APOE groups.

β values, 95% CIs, and P values result from linear regression models. β values are interpreted as the difference in the mean level in comparison to the APOE3 group. P values < 0.00013 were considered as statistically significant after applying a Bonferroni correction for multiple testing; statistically significant findings are shown in bold. Only immune activation markers with P values < 0.01 in adjusted analysis are shown. β, regression coefficient; CI, confidence interval.

APOE2 versus APOE3 (reference group) APOE4 versus APOE3 (reference group)
Unadjusted analysis Adjusting for age, sex, years of education, BMI, smoking, hypertension, diabetes, and dyslipidemia Unadjusted analysis Adjusting for age, sex, years of education, BMI, smoking, hypertension, diabetes, and dyslipidemia
β (95% CI) P value β (95% CI) P value β (95% CI) P value β (95% CI) P value
Established biomarkers
Aβ42 71.79 (−64.03, 207.61) 0.30 74.17 (−68.52, 216.86) 0.31 −280.02 (−396.07, −163.96) 3.44 × 10 −06 −279.43 (−398.48, −160.38) 6.22 × 10 −06
Total tau 14.43 (−19.62, 48.47) 0.41 11.39 (−20.87, 43.64) 0.49 25.87 (−3.98, 55.73) 0.089 34.13 (5.93, 62.34) 0.018
P-tau181 1.28 (−2.00, 4.56) 0.44 1.02 (−2.13, 4.17) 0.52 3.38 (0.45, 6.30) 0.024 4.14 (1.35, 6.94) 0.0038
NfL 0.02 (−0.18, 0.21) 0.88 0.05 (−0.13, 0.23) 0.59 0.09 (−0.08, 0.27) 0.30 0.15 (−0.02, 0.31) 0.093
Ng 0.13 (−0.10, 0.36) 0.28 0.08 (−0.15, 0.32) 0.49 0.10 (−0.09, 0.28) 0.31 0.14 (−0.04, 0.33) 0.13
SNAP25 0.10 (−0.05, 0.24) 0.19 0.09 (−0.06, 0.23) 0.24 0.34 (0.23, 0.46) 2.39 × 10 −08 0.38 (0.26, 0.49) 1.05 × 10 −09
YKL40 −11.00 (−47.87, 25.88) 0.56 −14.69 (−48.38, 19.00) 0.39 −1.73 (−31.13, 27.66) 0.91 13.12 (−13.60, 39.83) 0.33
Immune activation markers
CSF3 0.01 (−0.38, 0.40) 0.95 −0.04 (−0.44, 0.36) 0.85 −0.59 (−0.95, −0.23) 0.0014 −0.57 (−0.93, −0.21) 0.0023
FCRL3 −0.50 (−0.86, −0.14) 0.0067 −0.53 (−0.90, −0.16) 0.0056 −0.10 (−0.34, 0.15) 0.44 −0.10 (−0.36, 0.15) 0.43
IL-22RA1 −0.13 (−0.35, 0.09) 0.24 −0.14 (−0.37, 0.09) 0.24 −0.31 (−0.53, −0.09) 0.0052 −0.30 (−0.52, −0.08) 0.0084
IL-24 −0.06 (−0.18, 0.07) 0.38 −0.08 (−0.21, 0.05) 0.25 −0.15 (−0.25, −0.04) 0.0055 −0.14 (−0.25, −0.03) 0.0098
Open in a new tab

Fig. 1. Immune activation markers and AD biomarkers in CSF according to APOE group.

Fig. 1.

Open in a new tab

Violin plots of the median and quartiles for Aβ42 (A), SNAP25 (B), CSF3 (C), FCRL3 (D), IL-22RA1 (E), and IL-24 (F) are shown according to APOE groups. P values result from linear regression models that were adjusted for age, sex, years of education, BMI, smoking, hypertension, diabetes, and dyslipidemia.

Impact of age, sex, and other participant characteristics on AD biomarkers and immune activation markers in CSF among cognitively unimpaired participants

Age was the strongest factor influencing CSF AD biomarkers (Table 3 and table S2); older age was significantly associated with increased YKL40 (β: 68.21, P = 5.98 × 10−12), total tau (β: 52.23, P = 2.67 × 10−07), NfL (β: 0.30, P = 7.74 × 10−07), and p-tau181 (β: 4.73, P = 2.86 × 10−06) (Fig. 2), with suggestive associations identified with greater Ng (β: 0.23, P = 0.0005) and SNAP25 (β: 0.13, P = 0.0014). In addition, there were suggestive associations of higher NfL levels for males (β: 0.22, P = 0.0024) and between obesity and YKL40 (β: −33.39, P = 0.008) but no evident correlations with education, smoking, hypertension, diabetes, and dyslipidemia with CSF AD biomarkers in multivariable analysis. Age was the predominant influencer of CSF immune activation markers as evidenced by the smallest association P values in comparison to other clinical and epidemiological characteristics, where tumor necrosis factor receptor superfamily member 11B (TNFRSF11B), C-X-C motif chemokine ligand 9 (CXCL9), PDZ and LIM domain protein 7 (PDLIM7), secretoglobin family 1A member 1 (SCGB1A1), and galectin 9n (LGALS9) showed the strongest associations with age (Fig. 2). In addition, very strong associations with sex and diabetes were also observed for CSF immune activation markers, with males having higher levels of placental growth factor (PGF), matrix extracellular phosphoglycoprotein (MEPE), ephrin type-A receptor 1 (EPHA1), ectodysplasin A receptor (EDAR), and C-C motif chemokine ligand 23 (CCL23) (Table 3) and with participants with diabetes having higher levels of REG4 (regenerating family member 4), Galectin-4 (LGALS4), CCL21, fatty acid-binding protein 1 (FABP1), and trefoil factor 2 (TFF2) (table S2). We also observed a significant association between hypertension and SULT2A1 (table S2). When assessing interactions of age and sex with APOE group in regard to associations with established AD biomarkers and immune activation markers, no significant interactions were identified (tables S3 and S4).

Table 3. Associations of AD biomarkers and immune activation markers in CSF with age and sex.

β values, 95% CIs, and P values result from linear regression models. β values are interpreted as the change in the mean level corresponding to each 10-year increase in age or male sex. P values < 0.00013 were considered as statistically significant after applying a Bonferroni correction for multiple testing; statistically significant findings are shown in bold. The five immune activation markers with the smallest association P values are shown. β, regression coefficient; CI, confidence interval.

Unadjusted analysis Adjusting for sex, years of education, BMI, smoking, hypertension, diabetes, dyslipidemia, and APOE group
β (95% CI) P value β (95% CI) P value
Established biomarkers
Aβ42 24.79 (−52.94, 102.51) 0.53 17.84 (−62.74, 98.42) 0.66
Total tau 55.64 (37.24, 74.04) 7.85 × 10 −09 52.23 (32.75, 71.71) 2.67 × 10 −07
P-tau181 5.11 (3.26, 6.96) 1.15 × 10 −07 4.73 (2.78, 6.68) 2.86 × 10 −06
NfL 0.31 (0.20, 0.42) 1.16 × 10 −07 0.30 (0.18, 0.42) 7.74 × 10 −07
Ng 0.23 (0.11, 0.35) 0.0002 0.23 (0.10, 0.36) 0.0005
SNAP25 0.12 (0.04, 0.20) 0.0029 0.13 (0.05, 0.21) 0.0014
YKL40 72.27 (54.81, 89.73) 1.25 × 10 −14 68.21 (49.56, 86.87) 5.98 × 10 −12
Immune activation markers
TNFRSF11B 0.50 (0.41, 0.59) 1.00 × 10 −22 0.46 (0.37, 0.56) 1.55 × 10 −18
CXCL9 0.88 (0.71, 1.06) 2.50 × 10 −20 0.84 (0.65, 1.03) 2.20 × 10 −16
PDLIM7 0.44 (0.34, 0.53) 4.84 × 10 −17 0.43 (0.33, 0.53) 1.20 × 10 −14
SCGB1A1 0.90 (0.70, 1.10) 1.32 × 10 −16 0.84 (0.63, 1.04) 4.91 × 10 −14
LGALS9 0.26 (0.20, 0.32) 2.07 × 10 −15 0.25 (0.18, 0.31) 4.83 × 10 −13
Associations with sex
Established biomarkers
Aβ42 6.67 (−92.90, 106.25) 0.90 −33.82 (−133.1, 65.47) 0.50
Total tau 9.74 (−15.24, 34.71) 0.44 9.32 (−14.69, 33.32) 0.45
P-tau181 1.36 (−1.12, 3.85) 0.28 1.55 (−0.85, 3.95) 0.22
NfL 0.21 (0.06, 0.35) 0.0053 0.22 (0.08, 0.36) 0.0024
Ng −0.02 (−0.18, 0.14) 0.84 −0.05 (−0.21, 0.11) 0.52
SNAP25 0.05 (−0.05, 0.16) 0.32 0.07 (−0.03, 0.17) 0.16
YKL40 18.03 (−7.09, 43.16) 0.16 15.87 (−7.36, 39.10) 0.18
Immune activation markers
PGF 0.44 (0.33, 0.55) 3.42 × 10 −13 0.40 (0.30, 0.51) 1.23 × 10 −12
MEPE 0.70 (0.50, 0.90) 1.86 × 10 −11 0.64 (0.44, 0.84) 6.18 × 10 −10
EPHA1 0.45 (0.29, 0.61) 7.46 × 10 −08 0.39 (0.24, 0.55) 1.15 × 10 −06
EDAR 0.25 (0.16, 0.34) 7.75 × 10 −08 0.22 (0.13, 0.31) 1.90 × 10 −06
CCL23 0.43 (0.25, 0.60) 2.56 × 10 −06 0.38 (0.22, 0.55) 4.44 × 10 −06
Open in a new tab

Fig. 2. Associations of age with immune activation markers and AD biomarkers in CSF.

Fig. 2.

Open in a new tab

Scatterplots of age versus total tau (A), p-tau181 (B), NfL (C), YKL40 (D), TNFRSF11B (E), CXCL9 (F), PDLIM7 (G), SCGB1A1 (H), and LGALS9 (I) are shown with regression lines. Pearson correlation coefficient (r) and P values result from linear regression models that were adjusted for sex, years of education, BMI, smoking, hypertension, diabetes, dyslipidemia, and APOE group.

Associations between CSF immune activation markers and AD biomarkers in cognitively unimpaired participants

When examining associations between CSF immune activation markers and AD biomarkers, a substantial number (ranging from 64 to 172) of significant (P < 0.00013) associations were identified for each AD biomarker. Top hits were CXADR, CLSTN2, and BSG for Aβ42; DNAJA2, BSG, and TNFSF12 for total tau; DNAJA2, BSG, and TNFSF12 for p-tau181; FLT3LG, IL-15, and TNFRSF11A for NfL; ROBO1, VEGFA, and CRIM1 for SNAP25; BSG, DNAJA2, and CXADR for Ng; and BSG, DNAJA2, and CXADR for YKL40 (Table 4). When evaluating interactions of established AD biomarkers with APOE genotype in regard to associations with immune activation markers, TREM2 was found to interact with APOE2 for associations with CSF total tau (interaction P = 0.0001) and p-tau181 (interaction P = 2.54 × 10−05). There was also a significant interaction between FIS1 and APOE2 regarding the association with p-tau 181 (interaction P = 0.0001) (table S5). On the other hand, STX8 interacted with APOE4 for associations with both total tau (interaction P = 4.14 × 10−05) and p-tau181 (interaction P = 2.32 × 10−06) (table S6).

Table 4. Associations of CSF immune activation markers with AD biomarkers.

β values, 95% CIs, and P values result from linear regression models. β values are interpreted as the change in the mean level corresponding to each 500-unit increase in Aβ42, each 50-unit increase in total tau, each 10-unit increase in p-tau181, each doubling in NfL, each doubling in SNAP25, each doubling in Ng, and each 50-unit increase in YKL40. P values < 0.00013 were considered as statistically significant after applying a Bonferroni correction for multiple testing; statistically significant findings are shown in bold. The five immune activation markers with the smallest association P values are shown. β, regression coefficient; CI, confidence interval.

Unadjusted analysis Adjusting for age, sex, years of education, BMI, smoking, hypertension, diabetes, dyslipidemia, and APOE group
β (95% CI) P value β (95% CI) P value
Associations with Aβ42
CXADR 0.33 (0.25, 0.40) 2.33 × 10 −16 0.35 (0.27, 0.43) 2.15 × 10 −17
CLSTN2 0.19 (0.15, 0.24) 3.01 × 10 −16 0.21 (0.16, 0.25) 2.76 × 10 −17
BSG 0.24 (0.19, 0.30) 6.17 × 10 −16 0.26 (0.20, 0.32) 3.90 × 10 −17
CD200 0.22 (0.17, 0.27) 1.65 × 10 −15 0.23 (0.18, 0.29) 2.50 × 10 −16
IFNGR1 0.18 (0.14, 0.23) 2.68 × 10 −14 0.19 (0.15, 0.24) 9.88 × 10 −16
Associations with total tau
DNAJA2 0.18 (0.17, 0.20) 7.62 × 10 −73 0.18 (0.17, 0.20) 2.38 × 10 −65
BSG 0.17 (0.16, 0.19) 2.67 × 10 −68 0.18 (0.16, 0.20) 4.19 × 10 −64
TNFSF12 0.19 (0.18, 0.21) 1.09 × 10 −64 0.20 (0.18, 0.22) 5.83 × 10 −61
NELL2 −0.08 (−0.09, −0.07) 1.77 × 10 −50 −0.09 (−0.10, −0.08) 4.68 × 10 −54
CXADR 0.21 (0.19, 0.23) 3.47 × 10 −52 0.23 (0.20, 0.25) 4.94 × 10 −51
Associations with p-tau181
DNAJA2 0.34 (0.31, 0.38) 3.25 × 10 −59 0.34 (0.31, 0.38) 3.30 × 10 −53
BSG 0.32 (0.29, 0.36) 1.67 × 10 −53 0.33 (0.30, 0.37) 6.03 × 10 −50
TNFSF12 0.36 (0.32, 0.40) 1.06 × 10 −50 0.37 (0.33, 0.41) 1.77 × 10 −47
NELL2 −0.15 (−0.17, −0.13) 5.25 × 10 −40 −0.16 (−0.18, −0.14) 2.82 × 10 −42
DPP10 0.33 (0.29, 0.37) 2.60 × 10 −42 0.35 (0.31, 0.40) 2.38 × 10 −41
Associations with NfL
FLT3LG 0.17 (0.12, 0.21) 1.77 × 10 −12 0.13 (0.08, 0.17) 7.03 × 10 −08
IL-15 0.22 (0.16, 0.28) 4.63 × 10 −13 0.16 (0.10, 0.22) 1.08 × 10 −07
TNFRSF11A 0.28 (0.21, 0.36) 2.37 × 10 −12 0.22 (0.14, 0.30) 1.09 × 10 −07
CD276 0.28 (0.20, 0.35) 1.01 × 10 −12 0.20 (0.13, 0.28) 2.40 × 10 −07
OSCAR 0.27 (0.20, 0.34) 1.18 × 10 −12 0.20 (0.13, 0.27) 2.69 × 10 −07
Associations with SNAP25
ROBO1 0.59 (0.51, 0.66) 9.74 × 10 −38 0.67 (0.59, 0.74) 7.51 × 10 −45
VEGFA 0.51 (0.44, 0.58) 1.38 × 10 −34 0.57 (0.50, 0.64) 1.44 × 10 −41
CRIM1 0.39 (0.33, 0.44) 6.69 × 10 −34 0.43 (0.37, 0.48) 4.71 × 10 −38
FGF5 0.49 (0.41, 0.56) 4.45 × 10 −31 0.57 (0.49, 0.64) 2.17 × 10 −37
LRRN1 0.50 (0.42, 0.57) 5.48 × 10 −32 0.56 (0.49, 0.64) 2.39 × 10 −36
Associations with Ng
BSG 0.39 (0.36, 0.42) 6.89 × 10 −84 0.39 (0.37, 0.42) 9.39 × 10 −81
DNAJA2 0.40 (0.37, 0.43) 1.92 × 10 −80 0.39 (0.36, 0.42) 3.49 × 10 −78
CXADR 0.51 (0.47, 0.55) 1.77 × 10 −77 0.52 (0.48, 0.56) 2.46 × 10 −77
TNFSF12 0.44 (0.41, 0.48) 8.65 × 10 −81 0.45 (0.41, 0.48) 4.41 × 10 −77
DPP10 0.43 (0.40, 0.46) 6.05 × 10 −77 0.44 (0.40, 0.47) 4.01 × 10 −74
Associations with YKL40
BSG 0.13 (0.11, 0.15) 1.10 × 10 −32 0.14 (0.12, 0.16) 3.48 × 10 −28
DNAJA2 0.14 (0.12, 0.16) 6.05 × 10 −35 0.14 (0.12, 0.16) 1.05 × 10 −27
CXADR 0.17 (0.14, 0.19) 7.23 × 10 −29 0.18 (0.15, 0.21) 1.44 × 10 −27
IFNGR1 0.11 (0.09, 0.12) 4.02 × 10 −33 0.11 (0.09, 0.12) 7.32 × 10 −27
TNFSF12 0.15 (0.12, 0.17) 9.38 × 10 −30 0.15 (0.13, 0.18) 2.40 × 10 −26
Open in a new tab

Prediction for CDR conversion based on CSF immune activation markers and AD biomarkers

We investigated whether CSF immune activation markers and AD biomarkers at baseline in cognitively unimpaired participants, alone or in combination with participant characteristics, can predict CDR conversion from 0 to >0 in a subset of 286 who had longitudinal CDR measurements available. Of these participants, 83 (29.0%) experienced CDR conversion with a median follow-up length of 7.8 years (range: 1.0 to 16.5 years) after the baseline visit (table S7). There was a significant association between older age at the baseline visit and an increased risk of CDR conversion [hazard ratio (HR) = 2.85 (per 10-year increase), P = 1.11 × 10−07] in multivariable analysis; the c-index for age in predicting CDR conversion was 0.699. No significant or suggestive associations with CDR conversion were observed for other participant characteristics (Table 5). Among immune activation markers and AD biomarkers, a statistically significant association with CDR conversion in multivariable analysis adjusting for participant characteristics was observed for Aβ42 (HR = 0.63, P = 0.0001), with suggestive associations identified for NfL (HR = 1.34, P = 0.007), REG4 (HR = 1.43, P = 0.0009), MICB_MICA (HR = 0.73, P = 0.002), and SCGB3A2 (HR = 1.40, P = 0.009) with corresponding c-indexes of 0.617, 0.659, 0.630, 0.565, and 0.664, respectively (Table 5). Of note, the c-index for a model including the five AD biomarkers/immune activation markers that displayed significant or suggestive associations with CDR conversion in multivariable analysis, but not including any participant characteristics, was 0.754. To identify the strongest independent predictors of CDR conversion, we included all participant characteristics, Aβ42 and NfL, as well as REG4, MICB_MICA, and SCGB3A2, in a forward selection procedure in our Cox regression analysis. Statistically significant or suggestive associations were observed for age (HR = 2.14, P = 5.48 × 10−05), Aβ42 (HR = 0.53, P = 8.39 × 10−07), NfL (HR = 1.54, P = 4.52 × 10−05), and REG4 (HR = 1.35, P = 0.002). The model’s c-index was 0.762, which was a moderate improvement of 0.063 compared to an age only model (Fig. 3).

Table 5. Associations of participant characteristics, CSF immune activation markers, and AD biomarkers with CDR conversion.

HRs, 95% CIs, and P values result from Cox proportional hazards regression models. HRs are interpreted as the multiplicative increase in the risk of CDR conversion corresponding to each 1 SD increase in the given biomarker. Only immune activation markers with P values < 0.01 in adjusted analysis are shown. P values < 0.00013 were considered as statistically significant after applying a Bonferroni correction for multiple testing; statistically significant findings are shown in bold. HR, hazard ratio; CI, confidence interval.

Unadjusted analysis Adjusting for age, sex, years of education, BMI, smoking, hypertension, diabetes, dyslipidemia, and APOE group
C-index (95% CI) HR (95% CI) P value HR (95% CI) P value
Age (10-year increase) 0.699 (0.629, 0.770) 2.59 (1.84, 3.64) 4.14 × 10 −08 2.85 (1.94, 4.19) 1.11 × 10 −07
Sex (male) 0.527 (0.467, 0.587) 1.49 (0.94, 2.37) 0.088 1.51 (0.94, 2.40) 0.085
Education (5-year increase) 0.576 (0.506, 0.645) 0.71 (0.48, 1.06) 0.097 0.73 (0.49, 1.09) 0.12
BMI (5-unit increase) 0.550 (0.480, 0.621) 0.94 (0.74, 1.20) 0.63 0.97 (0.73, 1.28) 0.82
Smoking (former or current) 0.546 (0.485, 0.607) 1.11 (0.72, 1.70) 0.65 1.06 (0.68, 1.66) 0.80
Hypertension 0.556 (0.498, 0.614) 1.66 (0.99, 2.77) 0.053 1.13 (0.64, 2.00) 0.67
Diabetes 0.550 (0.503, 0.597) 1.68 (1.02, 2.76) 0.041 1.73 (0.99, 3.03) 0.053
Dyslipidemia 0.498 (0.452, 0.544) 1.03 (0.59, 1.82) 0.91 0.89 (0.47, 1.68) 0.72
APOE genotype 0.530 (0.470, 0.590) 1.20 (0.82, 1.74) 0.35 1.59 (1.04, 2.41) 0.031
Established biomarkers
Aβ42 0.617 (0.547, 0.687) 0.64 (0.51, 0.80) 0.0001 0.63 (0.49, 0.79) 0.0001
Total tau 0.578 (0.508, 0.649) 1.19 (0.96, 1.47) 0.11 0.96 (0.76, 1.22) 0.76
P-tau181 0.583 (0.513, 0.653) 1.22 (0.99, 1.49) 0.060 1.02 (0.81, 1.28) 0.87
NfL 0.659 (0.588, 0.731) 1.39 (1.16, 1.66) 0.0003 1.34 (1.08, 1.66) 0.0073
Ng 0.529 (0.457, 0.601) 1.13 (0.91, 1.42) 0.28 0.96 (0.77, 1.21) 0.73
SNAP25 0.531 (0.459, 0.603) 1.06 (0.85, 1.33) 0.59 0.82 (0.63, 1.07) 0.14
YKL40 0.618 (0.545, 0.690) 1.33 (1.07, 1.65) 0.0093 1.04 (0.81, 1.34) 0.77
Immune activation markers
REG4 0.630 (0.560, 0.701) 1.59 (1.33, 1.90) 3.93 × 10 −07 1.43 (1.16, 1.77) 0.0009
MICB_MICA 0.565 (0.494, 0.635) 0.79 (0.66, 0.96) 0.015 0.73 (0.60, 0.89) 0.0022
SCGB3A2 0.664 (0.594, 0.735) 1.74 (1.39, 2.18) 1.08 × 10 −06 1.40 (1.09, 1.81) 0.0092
Open in a new tab

Fig. 3. Cumulative incidence of CDR conversion after baseline visit.

Fig. 3.

Open in a new tab

Kaplan-Meier curves of the conversion from CDR = 0 to CDR > 0 after baseline visit according to APOE, age, and CSF levels of Aβ42, NfL, and REG4 are shown. The cohort was dichotomized by APOE group (A) or the median value of age (B), Aβ42 (C), NfL (D), and REG4 (E). (F) Kaplan-Meier curve of the CDR conversion was plotted on the basis of the number of risk factors (age > 73.9 years, Aβ42 ≤ 1197 pg/ml, NfL > 568.8 pg/ml, and REG4 > 0.0 NPX) for subjects without any missing data for any of these four variables.

For the validation, we analyzed datasets from the Charles F. and Joanne Knight Alzheimer’s Disease Research Center (Knight ADRC) at Washington University School of Medicine in St. Louis. Through the same inclusion criteria, 300 cognitively unimpaired participants (over 65 years of age, CDR = 0) with available APOE genotype, epidemiological/clinical characteristics, CSF biomarkers, and longitudinal CDR measurements available were investigated (table S8). Of these participants, 94 (31.3%) experienced CDR conversion with a median follow-up length of 5.2 years (range: 0.9 to 17.2 years) and a median CDR measurement number of 6 (range: 2 to 17). Multivariable analysis confirmed the association of CDR conversion with age [HR = 3.42 (per 10-year increase), P = 1.83 × 10−10] and CSF levels of Aβ42 (HR = 0.66, P = 0.0040) and NfL (HR = 1.43, P = 0.0067) but not REG4 (HR = 0.48, P = 0.41) (table S9).

DISCUSSION

The incidence of age-related dementia is influenced by multiple factors, although risk factors for 60% of cases remain unclear (17). In cognitively unimpaired individuals with a mean age in the mid-70s, our results show that APOE genotype and age significantly affect CSF levels of established AD biomarkers. As APOE4 is the strongest genetic risk factor for developing AD (18), we found that APOE4 correlates with lower CSF Aβ42 compared to APOE3, reflecting the presence of brain Aβ deposition (19, 20) even cognitively unimpaired. In addition, consistent with a previous report (21), increased CSF levels of presynaptic SNAP25, but not postsynaptic Ng and axonal NfL, were observed in cognitively unimpaired participants with APOE4. Significant effects of APOE4 on CSF Aβ42 and SNAP25 remained even after controlling for age, sex, and other factors, indicating that APOE4 predominantly exacerbates amyloid pathology and presynaptic damage. In contrast, age was associated with different domains of CSF AD biomarkers such as astrogliosis marker (YKL40), tauopathy marker (p-tau181), and axonal damage marker (total tau and NfL). Since there were no interactions between APOE and age on the CSF AD biomarker levels in the cognitively unimpaired participants, these factors may independently trigger the development of AD pathology, although further studies are needed in MCI and AD cohorts. While sex, education, obesity, smoking, hypertension, diabetes, and dyslipidemia are involved in AD and dementia risk (1, 17), significant associations with the CSF AD biomarker levels were not detected in our study. Thus, these risk factors may influence AD pathology in later stages of the disease and/or aggravate clinical symptoms caused by AD pathology.

In addition to AD pathology related to ATN classification, sustained brain inflammation has emerged as a central hallmark of AD (22). Our proximity extension assay revealed that age is a factor which strongly affects CSF immune activation marker levels in cognitively unimpaired individuals. Aging causes immune system remodeling (23–25), where CXCL9 (C-X-C motif chemokine ligand 9) has been shown as the strongest contributor to age-related chronic inflammation in blood samples (26). Our study also found a strong association between age and CXCL9 in CSF. Since CXCL9 contributes to arterial stiffness by leading to endothelial cell senescence (26), increased CXCL9 in CSF may represent age-related cerebrovascular damage and the presence of cerebral small vessel disease. While APOE effects on CSF immune activation markers were not evident, CSF3 (colony-stimulating factor 3), also known as GCSF (granulocyte colony-stimulating factor), was lower in APOE4 subjects than APOE3 subjects. CSF3 is an essential growth factor regulating the maturation of granulocytes (27). Furthermore, CSF3 is known to stimulate neurogenesis (28) and has neuroprotective functions (29). The decreased CSF3 levels may partially contribute to APOE4’s risk to AD. Since CSF3 administration ameliorates AD-related phenotypes in mouse models (30–32), a clinical trial has been performed using CSF3 to treat patients with AD (33). Although the clinical trial did not yield evident beneficial results in the patients unlike preclinical studies, our finding suggests that GCSF therapy could potentially be more beneficial in treatment of AD patients with APOE4.

Accumulating evidence indicates that a maladaptive immune system initiates the transition from preclinical stage to MCI and dementia in AD progression (34). Since many CSF immune activation marker levels highly correlate with AD biomarkers even in cognitively unimpaired individuals, they may be able to detect subtle AD pathological changes. We found several immune activation markers: BSG (Basigin), DNAJA2 (DnaJ homolog subfamily A member 2), and TNFSF12 (TNF superfamily member 12) all had strong associations with the set of AD biomarkers except for NfL and SNAP25. BSG, known as CD147, is a multifunctional transmembrane glycoprotein. By interacting with various ligands on the cell surface, BSG mediates immune responses and neuronal/glial homeostasis (35). Specifically, BSG is a regulatory subunit of γ-secretase, where BSG inhibition facilities Aβ production (36). Furthermore, BSG has been implicated in blood-brain barrier (BBB) degeneration through matrix metallopeptidase 9 (MMP9) activation (37, 38). As BBB dysfunction and leakage are present years before AD onset and affect neuronal dysfunction (39–41), the association between BSG and AD biomarker levels may be a consequence of cerebrovascular damage. Tau assembly in tauopathies is mediated by various chaperones including HSP70 and HSP90. DNAJA2 belongs to the HSP40 family and is the strongest inhibitor of tau aggregation (42). Specifically, DNAJA2 is detected in p-tau–positive neurons in MCI and AD postmortem brains (42). Consistent with the report, we identified DNAJA2 as the top-ranked molecule associated with total tau and p-tau181, suggesting that DNAJA2 may be a potential CSF biomarker for tauopathy. TNFSF12, also known as TWEAK (TNF-like weak inducer of apoptosis), functions as a multifunctional proinflammatory and pro-angiogenic cytokine as well as an apoptosis inducer (43). Since TNFSF12 likely mediates the process of tissue repair upon acute injury (43), TNFSF12 may be up-regulated with significant associations with AD biomarker levels in response to presymptomatic neuronal damage. Further studies should define whether these immune activation markers are causatively or consequently involved in pathogenic mechanisms for age-related cognitive decline and AD.

Our results provide evidence that age and CSF Aβ42, and possibly CSF NfL and REG4, are independently associated with CDR conversion risk in cognitively unimpaired individuals. As expected, age displayed the strongest association with risk of CDR conversion. Combining the identified CSF biomarkers (Aβ42, NfL, and REG4) with age can improve conversion prediction compared to age alone in our study. When dichotomizing these risk factors by their median values, of the 28 subjects with old age, low Aβ42, high NfL, and high REG4 at baseline visit, 78.8% had developed cognitive impairment during the follow-up for 10 years, compared to 5.9% of the 19 subjects with none of the four risk factors. While reports show the association of CSF Aβ42 (44) and NfL (45) with CDR conversion, we demonstrate that REG4 is an independent CSF biomarker capable of predicting age-related cognitive impairment. REG4 is most expressed in the colon and small intestine (46, 47) and up-regulated in inflammatory bowel disease or ulcerative colitis (48). REG family proteins, including REG3α, REG3β, REG3γ, and REG4, likely regulate the gut microbiome, thereby contributing to intestinal inflammation (49). Critical roles of gut-brain axis to neurodegenerative disease pathogenesis have been increasingly recognized (50, 51). Evidence indicates that gut microbiome composition changes are associated with AD risk (52). Thus, our results support that altering the gut microbiome affects the likelihood of cognitive decline, where the REG4 level in CSF may reflect pathway progression. Furthermore, we found that REG4 is strongly associated with obesity and diabetes. Obesity and diabetes also display an increased risk in developing vascular pathologies, increasing inflammation, and altering lipid metabolism, all of which are risk factors for AD (53, 54). However, since no evident association between CDR conversion and CSF REG4 levels measured by the aptamer-based SomaScan platform was observed in the replication cohort, further research is required to disambiguate how REG4 is involved in the metabolomic inflammation as well as bidirectional inflammatory communication along the gut-brain axis.

Numerous studies have investigated CSF proteome profiling using the Olink proximity extension assay in MCI and AD cases (55, 56). Among the 85 immune activation markers associated with CDR conversion in our cohort with P values < 0.05 in unadjusted analysis, CSF levels of WNT9A, IL-6, and CXCL6 were altered in MCI cases with amyloid pathology compared to control groups (55). While higher CSF levels of MMP10, TNFSF13, ANGPTL4, LTBR, WNT9A, CD4, FLT3LG, CD84, MICB_MICA, and CCL23 were detected in AD cases than control groups in the 85 immune activation markers (55), another study also found the increases of MMP10, TNFSF13, and CD4 in CSF from autosomal dominant AD (56).These observations indicate that some CSF immune activation markers start changing even at presymptomatic early stages of cognitive impairment. REG4 was recognized as the top-ranked CSF immune activation marker predicting CDR conversion independently of other factors in cognitively unimpaired elderly. However, there were no apparent differences in CSF REG4 levels between control and MCI/AD subjects in other cohorts (55, 56). CSF REG4 was up-regulated in AD compared to non-AD dementia cases including dementia with Lewy bodies and frontotemporal dementia (55). Elevated levels of CSF REG4 in early-stage asymptomatic AD subjects among the control groups may mask the differences between control and MCI/AD. However, further studies are necessary to explore this in depth. In addition, we found a positive correlation between CSF TREM2 levels and CDR conversion risk (HR = 1.48, P = 0.0020) in the unadjusted analysis. In contrast, Alzheimer’s Disease Neuroimaging Initiative datasets revealed that higher CSF sTREM2 levels were associated with slower cognitive and clinical decline in AD (57), along and with a lower risk of AD conversion in amyloid positron emission tomography–positive MCI cases (58). Thus, TREM2-related immune activation may contribute differently to the progression of age-related cognitive decline and AD symptoms depending on the disease status and amyloid pathology. The Olink proximity extension assay also demonstrated the association of plasma immune activation markers with cognitive function and dementia (59). Since executive function was negatively associated with plasma IL-6, IL-7, and TGFB1 levels (59) among the 85 immune activation markers, our future study should determine whether plasma immune activation markers can predict the CDR conversion in our cohort.

In conclusion, our study demonstrated that CSF immune activation markers strongly correlate with AD biomarkers. Combining CSF biomarkers and epidemiological/clinical characteristics can better clarify a propensity to develop cognitive decline before symptom onset in the elderly. Future studies should validate our findings in independent cohorts. However, our study was conducted in the MCSA, which is from a geographically defined population with relatively less diversity. Insights drawn from the dataset may be overstated. In addition, the possibility of a false-negative finding after correcting for multiple testing in the relatively small sample size should be carefully considered. Nonetheless, our findings might allow researchers and health care professionals to better understand the processes of age-related cognitive decline and identify potential therapeutic targets and individualized medicine.

MATERIALS AND METHODS

Study design

CSF samples from cognitively unimpaired individuals enrolled in the MCSA cohort (15) were analyzed in our study. The MCSA includes participants from Olmsted County, Minnesota in the United States. The participants underwent a battery of examinations: a neurological evaluation, risk factor ascertainment, neuropsychological testing, and self- and informant reported neuropsychiatric symptoms. Participants are evaluated every 15 months following the same protocol. The MCSA has been approved by the Mayo Clinic Institutional Review Boards and Olmsted Medical Center. This study follows Health Insurance Portability and Accountability Act guidelines and obtained written informed consent from all participants. Our cohort included 298 cognitively unimpaired participants (CDR global score = 0, >65 years old) with available CSF samples and APOE genotypes. All subjects were unrelated, non-Hispanic/Latino, and white. Among the cohort, longitudinal CDR conversion analysis included 286 cases with CDR-SB = 0 at the baseline visit and at least one follow-up visit for cognitive testing. The median number of CDR measurements was 7 (range: 2 to 13 visits). CDR conversion was defined as the occurrence of the first CDR-SB greater than 0. As a replication cohort, the dataset of 300 cognitively unimpaired participants from Knight ADRC identified through the same inclusion criteria was used for analyzed longitudinal CDR conversion analysis.

CSF measurements for AD biomarkers and immune activation markers

CSF samples were collected through lumbar puncture and centrifuged at 2000g for 10 min to remove leukocytes. To avoid potential blood contamination, the first 1 to 2 ml of CSF was not used for the biomarker measurements. Datasets for CSF Aβ42, total tau, and p-tau181 levels measured by automated electrochemiluminescence Elecsys immunoassays (Roche Diagnostics) and for CSF NfL levels measured by enzyme-linked immunosorbent assay (ELISA) were obtained from MCSA (45). CSF Ng and SNAP-25 levels were measured by microparticle-based immunoassays using Single Molecule Counting technology originally developed for the Erenna System by Singulex and now part of EMD Millipore (Burlington, MA) (6, 7). CSF YKL-40 level was measured by ELISA (Quidel MicroVue Bone, San Diego, CA) (8). CSF immune activation markers including 365 immune-related proteins were measured by proximity extension assay using the Olink human inflammation panel. The proximity extension assay uses protein-specific antibodies conjugated to oligonucleotide tags. Epitope-specific binding causes paired oligonucleotide DNA tags to hybridize. The samples are subsequently run through quantitative polymerase chain reaction to generate a log base-2 normalized protein expression (NPX) value (60). The median proportion of missing data across all 372 biomarkers was 0.0% (range: 0.0 to 11.1%). For the CDR conversion subcohort, the median proportion of missing data across all biomarkers was 0.0% (range: 0.0 to 10.8%). All data from Olink were used regardless of the lower limit of quantitation denotation. In the replication cohort, datasets for CSF Aβ42, total tau, and p-tau181 levels measured by ELISA (Innotest; Innogenetics, Ghent, Belgium) (19) and for CSF NfL levels measured by ELISA (UmanDiagnostics, Umeå, Sweden) (61) were obtained from Knight ADRC. CSF REG4 levels were measured using the SomaScan platform (SomaLogic, Boulder, CO) (62).

Statistical analysis

Continuous variables were summarized with the sample median and range. Categorical variables were summarized with number and percentage of subjects. Comparisons of participant characteristics according to APOE genotype group were made using a Kruskal-Wallis rank sum test (continuous variables) or Fisher’s exact test (categorical variables).

Associations of established AD biomarkers and immune activation markers with APOE genotype group (APOE2 versus APOE3, APOE4 versus APOE3, and ordinal APOE group), age, sex, years of education, obesity, smoking, hypertension, diabetes, and dyslipidemia were evaluated using unadjusted and multivariable linear regression models. Multivariable models were adjusted for APOE genotype group, age, sex, years of education, obesity, smoking, hypertension, diabetes, and dyslipidemia. Regression coefficients (denoted as β) and 95% confidence intervals (CIs) were estimated and are interpreted as the change in the mean biomarker level in comparison to the reference group (categorical variables) or corresponding to a specified increase (continuous variables). NfL, Ng, and SNAP25 were examined on the logarithm scale in all regression analyses due to their skewed distributions.

Associations of established AD biomarkers with immune activation markers were evaluated using the previously described unadjusted and multivariable linear regression models. When assessing interactions of APOE genotype group with clinical covariates (age, sex, years of education, obesity, smoking, hypertension, diabetes, and dyslipidemia) and established AD biomarkers with regard to associations with AD biomarkers and immune activation markers, the aforementioned multivariable linear regression models were used with the addition of an interaction term between APOE genotype group and the given clinical covariate/established biomarker of interest.

The Kaplan-Meier method was used to estimate the cumulative incidence of CDR conversion after the baseline visit, where censoring occurred at the date of the final visit. Associations of participant characteristics, AD biomarkers, and CSF immune activation markers with CDR conversion were evaluated using unadjusted and multivariable Cox proportional hazards regression models. Multivariable models were adjusted for APOE genotype group, age, sex, years of education, obesity, smoking, hypertension, diabetes, and dyslipidemia. HRs and 95% CIs were estimated. For AD biomarkers and CSF immune activation markers, HRs correspond to each 1 SD increase. NFL, Ng, and SNAP25 were again examined on the logarithm scale owing to their skewed distributions.

To assess the ability of clinical characteristics, established AD biomarkers, and immune activation markers to accurately predict CDR conversion, we first used unadjusted Cox proportional hazards regression models in an exploratory analysis, where of primary interest was estimation of the model c-index; a c-index of 1.0 indicates perfect predictive ability, and a c-index of 0.5 indicates predictive ability equal to chance. Subsequently, in the primary analysis, we used multivariable Cox proportional hazards regression models, applying a forward selection approach for variable selection with a P value < 0.01 criteria for model entry. Variables that were considered in forward selection included all clinical characteristics, as well as AD biomarkers and immune activation markers that were associated with CDR conversion with a P value < 0.01 when adjusting for all clinical characteristics; c-index was again estimated.

A Bonferroni correction for multiple testing was used to account for the 372 different AD biomarkers/immune activation markers that were assessed in association analysis, after which P values < 0.00013 were considered statistically significant. However, since applying a correction for multiple testing limits the probability of a type I error (i.e., false-positive finding), it correspondingly increases the likelihood of a type II error (i.e., a false-negative finding), and therefore, P values < 0.01 were considered as displaying suggestive evidence of an association. All statistical tests were two-sided. Statistical analyses were performed using SAS (version 9.4; SAS Institute Inc., Cary, NC).

Acknowledgments

We gratefully thank all the participants and their families. We acknowledge the contributions of the staff in Administration, Clinical, Biomarker, and Biostatistics Cores at MCSA and Mayo Clinic Alzheimer’s Disease Research Center.

Funding: This work was supported by National Institutes of Health grant U19 AG069701 (D.M.H. and T.K.), National Institutes of Health grant RF1 AG071226 (T.K.), National Institutes of Health grant RF1 AG057181 (T.K.), National Institutes of Health grant RF1 AG068034 (T.K.), National Institutes of Health grant RF1 AG081203 (T.K.), National Institutes of Health grant U01 AG006786 (R.C.P.), National Institutes of Health grant P30 AG062677 (R.C.P.), National Institutes of Health grant R01 AG044546 (C.C.), National Institutes of Health grant RF1 AG053303 (C.C.), National Institutes of Health grant RF1 AG058501 (C.C.), National Institutes of Health grant U01 AG058922 (C.C.), National Institutes of Health grant RF1 AG074007 (Y.J.S.), Alzheimer’s Association Zenith Fellows Award ZEN-22-848604 (C.C.), and Chan Zuckerberg Initiative grant (C.C.).

Author contributions: Conceptualization: D.M.H., G.B., M.G.H., and T.K. Investigation: F.S., L.J.W., R.H., J.U., R.L.H., W.K., Y.A.M., N.W., B.R., S.C.S., Y.R., C.X., Y.W.A., M.G.H., and T.K. Data generation: J.T., Y.J.S., and C.C. Visualization: F.S., L.J.W., M.G.H., and T.K. Project administration: J.A.S., M.V., M.M.M., and R.C.P. Supervision: D.M.H., G.B., M.G.H., and T.K. Writing—original draft: F.S., M.G.H., and T.K. Writing—review and editing: All.

Competing interests: R.C.P. is a consultant for Roche Inc., Genentech Inc., Eli Lilly and Co., Eisai Inc., and Nestle Inc. R.C.P. receives royalties from Oxford University Press and UpToDate and educational materials for Medscape. D.M.H. cofounded and is on the scientific advisory board of C2N Diagnostics. D.M.H. is on the scientific advisory board of Denali, Cajal Neuroscience, and Genentech and consults for Asteroid Therapeutics. C.C. has received research support from GSK and EISAI. C.C. is a member of the advisory board of Circular Genomics and owns stocks. M.M.M. has served on scientific advisory boards and/or has consulted for Biogen, Eisai, LabCorp, Lilly, Merck, PeerView Institute, Roche, Siemens Healthineers, and Sunbird Bio. M.V. has received research funding from F. Hoffmann-La Roche Ltd. and Biogen in the past, consulted for F. Hoffmann-La Roche Ltd., and has equity ownership in Johnson and Johnson, Merck, Medtronic, and Amgen. R.H. is a current employee of Life Molecular Imaging. Y.A.M. is currently an employee of SciNeuro Pharmaceuticals. All other authors declare that they have no competing interests.

Data and materials availability: All data needed to evaluate the conclusions in the paper are present in the paper and/or the Supplementary Materials.

Supplementary Materials

This PDF file includes:

Tables S1 to S9

Legends for data S1 to S8

sciadv.adk3674_sm.pdf (423.4KB, pdf)

Other Supplementary Material for this manuscript includes the following:

Data S1 to S8

sciadv.adk3674_data_s1_to_s8.zip (2.9MB, zip)

REFERENCES AND NOTES

  • 1.2023 Alzheimer's disease facts and figures. Alzheimers Dement. 19, 1598–1695 (2023). [DOI] [PubMed] [Google Scholar]
  • 2.Jack C. R. Jr., Bennett D. A., Blennow K., Carrillo M. C., Dunn B., Haeberlein S. B., Holtzman D. M., Jagust W., Jessen F., Karlawish J., Liu E., Molinuevo J. L., Montine T., Phelps C., Rankin K. P., Rowe C. C., Scheltens P., Siemers E., Snyder H. M., Sperling R., NIA-AA research framework: Toward a biological definition of alzheimer's disease. Alzheimers Dement. 14, 535–562 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Olsson B., Lautner R., Andreasson U., Ohrfelt A., Portelius E., Bjerke M., Holtta M., Rosen C., Olsson C., Strobel G., Wu E., Dakin K., Petzold M., Blennow K., Zetterberg H., CSF and blood biomarkers for the diagnosis of Alzheimer's disease: A systematic review and meta-analysis. Lancet Neurol. 15, 673–684 (2016). [DOI] [PubMed] [Google Scholar]
  • 4.Zetterberg H., Bendlin B. B., Biomarkers for Alzheimer's disease-preparing for a new era of disease-modifying therapies. Mol. Psychiatry 26, 296–308 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Park S. A., Han S. M., Kim C. E., New fluid biomarkers tracking non-amyloid-β and non-tau pathology in Alzheimer's disease. Exp. Mol. Med. 52, 556–568 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Schindler S. E., Li Y., Todd K. W., Herries E. M., Henson R. L., Gray J. D., Wang G., Graham D. L., Shaw L. M., Trojanowski J. Q., Hassenstab J. J., Benzinger T. L. S., Cruchaga C., Jucker M., Levin J., Chhatwal J. P., Noble J. M., Ringman J. M., Graff-Radford N. R., Holtzman D. M., Ladenson J. H., Morris J. C., Bateman R. J., Xiong C., Fagan A. M.; Dominantly Inherited Alzheimer Network , Emerging cerebrospinal fluid biomarkers in autosomal dominant Alzheimer's disease. Alzheimers Dement. 15, 655–665 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Sutphen C. L., McCue L., Herries E. M., Xiong C., Ladenson J. H., Holtzman D. M., Fagan A. M., Longitudinal decreases in multiple cerebrospinal fluid biomarkers of neuronal injury in symptomatic late onset Alzheimer's disease. Alzheimers Dement. 14, 869–879 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Sutphen C. L., Jasielec M. S., Shah A. R., Macy E. M., Xiong C., Vlassenko A. G., Benzinger T. L., Stoops E. E., Vanderstichele H. M., Brix B., Darby H. D., Vandijck M. L., Ladenson J. H., Morris J. C., Holtzman D. M., Fagan A. M., Longitudinal cerebrospinal fluid biomarker changes in preclinical alzheimer disease during middle age. JAMA Neurol. 72, 1029–1042 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Dubois B., Villain N., Frisoni G. B., Rabinovici G. D., Sabbagh M., Cappa S., Bejanin A., Bombois S., Epelbaum S., Teichmann M., Habert M. O., Nordberg A., Blennow K., Galasko D., Stern Y., Rowe C. C., Salloway S., Schneider L. S., Cummings J. L., Feldman H. H., Clinical diagnosis of Alzheimer's disease: Recommendations of the international working group. Lancet Neurol. 20, 484–496 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Hu W. T., Ozturk T., Kollhoff A., Wharton W., Christina Howell J.; Alzheimer’s Disease Neuroimaging Initiative , Higher CSF sTNFR1-related proteins associate with better prognosis in very early Alzheimer's disease. Nat. Commun. 12, 4001 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Hellwig K., Kvartsberg H., Portelius E., Andreasson U., Oberstein T. J., Lewczuk P., Blennow K., Kornhuber J., Maler J. M., Zetterberg H., Spitzer P., Neurogranin and YKL-40: Independent markers of synaptic degeneration and neuroinflammation in Alzheimer's disease. Alzheimers Res. Ther. 7, 74 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Janelidze S., Mattsson N., Stomrud E., Lindberg O., Palmqvist S., Zetterberg H., Blennow K., Hansson O., CSF biomarkers of neuroinflammation and cerebrovascular dysfunction in early Alzheimer disease. Neurology 91, e867–e877 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Suarez-Calvet M., Araque Caballero M. A., Kleinberger G., Bateman R. J., Fagan A. M., Morris J. C., Levin J., Danek A., Ewers M., Haass C.; Dominantly Inherited Alzheimer Network , Early changes in CSF sTREM2 in dominantly inherited Alzheimer's disease occur after amyloid deposition and neuronal injury. Sci. Transl. Med. 8, 369ra178 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Suarez-Calvet M., Kleinberger G., Araque Caballero M. A., Brendel M., Rominger A., Alcolea D., Fortea J., Lleo A., Blesa R., Gispert J. D., Sanchez-Valle R., Antonell A., Rami L., Molinuevo J. L., Brosseron F., Traschutz A., Heneka M. T., Struyfs H., Engelborghs S., Sleegers K., Van Broeckhoven C., Zetterberg H., Nellgard B., Blennow K., Crispin A., Ewers M., Haass C., sTREM2 cerebrospinal fluid levels are a potential biomarker for microglia activity in early-stage Alzheimer's disease and associate with neuronal injury markers. EMBO Mol. Med. 8, 466–476 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Roberts R. O., Geda Y. E., Knopman D. S., Cha R. H., Pankratz V. S., Boeve B. F., Ivnik R. J., Tangalos E. G., Petersen R. C., Rocca W. A., The mayo clinic study of aging: Design and sampling, participation, baseline measures and sample characteristics. Neuroepidemiology 30, 58–69 (2008). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Morris J. C., The Clinical Dementia Rating (CDR): Current version and scoring rules. Neurology 43, 2412–2414 (1993). [DOI] [PubMed] [Google Scholar]
  • 17.Livingston G., Huntley J., Sommerlad A., Ames D., Ballard C., Banerjee S., Brayne C., Burns A., Cohen-Mansfield J., Cooper C., Costafreda S. G., Dias A., Fox N., Gitlin L. N., Howard R., Kales H. C., Kivimaki M., Larson E. B., Ogunniyi A., Orgeta V., Ritchie K., Rockwood K., Sampson E. L., Samus Q., Schneider L. S., Selbaek G., Teri L., Mukadam N., Dementia prevention, intervention, and care: 2020 report of the Lancet Commission. Lancet 396, 413–446 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Yamazaki Y., Zhao N., Caulfield T. R., Liu C. C., Bu G., Apolipoprotein E and Alzheimer disease: Pathobiology and targeting strategies. Nat. Rev. Neurol. 15, 501–518 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Fagan A. M., Head D., Shah A. R., Marcus D., Mintun M., Morris J. C., Holtzman D. M., Decreased cerebrospinal fluid Abeta(42) correlates with brain atrophy in cognitively normal elderly. Ann. Neurol. 65, 176–183 (2009). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Baker J. E., Lim Y. Y., Pietrzak R. H., Hassenstab J., Snyder P. J., Masters C. L., Maruff P., Cognitive impairment and decline in cognitively normal older adults with high amyloid-β: A meta-analysis. Alzheimers Dement. (Amst) 6, 108–121 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Butt O. H., Long J. M., Henson R. L., Herries E., Sutphen C. L., Fagan A. M., Cruchaga C., Ladenson J. H., Holtzman D. M., Morris J. C., Ances B. M., Schindler S. E.; Alzheimer’s Disease Neuroimaging Initiative , Cognitively normal APOE ε4 carriers have specific elevation of CSF SNAP-25. Neurobiol. Aging 102, 64–72 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Kinney J. W., Bemiller S. M., Murtishaw A. S., Leisgang A. M., Salazar A. M., Lamb B. T., Inflammation as a central mechanism in Alzheimer's disease. Alzheimers Dement. (N Y) 4, 575–590 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Goldberg E. L., Dixit V. D., Drivers of age-related inflammation and strategies for healthspan extension. Immunol. Rev. 265, 63–74 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.Chung H. Y., Kim D. H., Lee E. K., Chung K. W., Chung S., Lee B., Seo A. Y., Chung J. H., Jung Y. S., Im E., Lee J., Kim N. D., Choi Y. J., Im D. S., Yu B. P., Redefining chronic inflammation in aging and age-related diseases: Proposal of the senoinflammation concept. Aging Dis. 10, 367–382 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.Rea I. M., Gibson D. S., McGilligan V., McNerlan S. E., Alexander H. D., Ross O. A., Age and age-related diseases: Role of inflammation triggers and cytokines. Front. Immunol. 9, 586 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Sayed N., Huang Y., Nguyen K., Krejciova-Rajaniemi Z., Grawe A. P., Gao T., Tibshirani R., Hastie T., Alpert A., Cui L., Kuznetsova T., Rosenberg-Hasson Y., Ostan R., Monti D., Lehallier B., Shen-Orr S. S., Maecker H. T., Dekker C. L., Wyss-Coray T., Franceschi C., Jojic V., Haddad F., Montoya J. G., Wu J. C., Davis M. M., Furman D., An inflammatory aging clock (iAge) based on deep learning tracks multimorbidity, immunosenescence, frailty and cardiovascular aging. Nat Aging 1, 598–615 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Mehta H. M., Corey S. J., G-CSF, the guardian of granulopoiesis. Semin. Immunol. 54, 101515 (2021). [DOI] [PubMed] [Google Scholar]
  • 28.Jung K. H., Chu K., Lee S. T., Kim S. J., Sinn D. I., Kim S. U., Kim M., Roh J. K., Granulocyte colony-stimulating factor stimulates neurogenesis via vascular endothelial growth factor with STAT activation. Brain Res. 1073-1074, 190–201 (2006). [DOI] [PubMed] [Google Scholar]
  • 29.Solaroglu I., Jadhav V., Zhang J. H., Neuroprotective effect of granulocyte-colony stimulating factor. Front. Biosci. 12, 712–724 (2007). [DOI] [PubMed] [Google Scholar]
  • 30.Sanchez-Ramos J., Song S., Sava V., Catlow B., Lin X., Mori T., Cao C., Arendash G. W., Granulocyte colony stimulating factor decreases brain amyloid burden and reverses cognitive impairment in Alzheimer's mice. Neuroscience 163, 55–72 (2009). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Prakash A., Medhi B., Chopra K., Granulocyte colony stimulating factor (GCSF) improves memory and neurobehavior in an amyloid-β induced experimental model of Alzheimer's disease. Pharmacol. Biochem. Behav. 110, 46–57 (2013). [DOI] [PubMed] [Google Scholar]
  • 32.Tsai K. J., Tsai Y. C., Shen C. K., G-CSF rescues the memory impairment of animal models of Alzheimer's disease. J. Exp. Med. 204, 1273–1280 (2007). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Sanchez-Ramos J., Cimino C., Avila R., Rowe A., Chen R., Whelan G., Lin X., Cao C., Ashok R., Pilot study of granulocyte-colony stimulating factor for treatment of Alzheimer's disease. J. Alzheimers Dis. 31, 843–855 (2012). [DOI] [PubMed] [Google Scholar]
  • 34.Simons M., Levin J., Dichgans M., Tipping points in neurodegeneration. Neuron 111, 2954–2968 (2023). [DOI] [PubMed] [Google Scholar]
  • 35.Muramatsu T., Basigin (CD147), a multifunctional transmembrane glycoprotein with various binding partners. J. Biochem. 159, 481–490 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Zhou S., Zhou H., Walian P. J., Jap B. K., CD147 is a regulatory subunit of the gamma-secretase complex in Alzheimer's disease amyloid beta-peptide production. Proc. Natl. Acad. Sci. U.S.A. 102, 7499–7504 (2005). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.Zhu X., Song Z., Zhang S., Nanda A., Li G., CD147: A novel modulator of inflammatory and immune disorders. Curr. Med. Chem. 21, 2138–2145 (2014). [DOI] [PubMed] [Google Scholar]
  • 38.Wang H., Lv J. J., Zhao Y., Wei H. L., Zhang T. J., Yang H. J., Chen Z. N., Jiang J. L., Endothelial genetic deletion of CD147 induces changes in the dual function of the blood-brain barrier and is implicated in Alzheimer's disease. CNS Neurosci. Ther. 27, 1048–1063 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39.Zlokovic B. V., Neurovascular pathways to neurodegeneration in Alzheimer's disease and other disorders. Nat. Rev. Neurosci. 12, 723–738 (2011). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40.Sweeney M. D., Sagare A. P., Zlokovic B. V., Blood-brain barrier breakdown in Alzheimer disease and other neurodegenerative disorders. Nat. Rev. Neurol. 14, 133–150 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 41.Sweeney M. D., Zhao Z., Montagne A., Nelson A. R., Zlokovic B. V., Blood-brain barrier: From physiology to disease and back. Physiol. Rev. 99, 21–78 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 42.Mok S. A., Condello C., Freilich R., Gillies A., Arhar T., Oroz J., Kadavath H., Julien O., Assimon V. A., Rauch J. N., Dunyak B. M., Lee J., Tsai F. T. F., Wilson M. R., Zweckstetter M., Dickey C. A., Gestwicki J. E., Mapping interactions with the chaperone network reveals factors that protect against tau aggregation. Nat. Struct. Mol. Biol. 25, 384–393 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43.Winkles J. A., The TWEAK-Fn14 cytokine-receptor axis: Discovery, biology and therapeutic targeting. Nat. Rev. Drug Discov. 7, 411–425 (2008). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 44.Roe C. M., Ances B. M., Head D., Babulal G. M., Stout S. H., Grant E. A., Hassenstab J., Xiong C., Holtzman D. M., Benzinger T. L. S., Schindler S. E., Fagan A. M., Morris J. C., Incident cognitive impairment: Longitudinal changes in molecular, structural and cognitive biomarkers. Brain 141, 3233–3248 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 45.Kern S., Syrjanen J. A., Blennow K., Zetterberg H., Skoog I., Waern M., Hagen C. E., van Harten A. C., Knopman D. S., Jack C. R. Jr., Petersen R. C., Mielke M. M., Association of cerebrospinal fluid neurofilament light protein with risk of mild cognitive impairment among individuals without cognitive impairment. JAMA Neurol. 76, 187–193 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 46.Kämäräinen M., Heiskala K., Knuutila S., Heiskala M., Winqvist O., Andersson L. C., RELP, a novel human REG-like protein with up-regulated expression in inflammatory and metaplastic gastrointestinal mucosa. Am. J. Pathol. 163, 11–20 (2003). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 47.Oue N., Mitani Y., Aung P. P., Sakakura C., Takeshima Y., Kaneko M., Noguchi T., Nakayama H., Yasui W., Expression and localization of Reg IV in human neoplastic and non-neoplastic tissues: Reg IV expression is associated with intestinal and neuroendocrine differentiation in gastric adenocarcinoma. J. Pathol. 207, 185–198 (2005). [DOI] [PubMed] [Google Scholar]
  • 48.Edwards J. A., Tan N., Toussaint N., Ou P., Mueller C., Stanek A., Zinsou V., Roudnitsky S., Sagal M., Dresner L., Schwartzman A., Huan C., Role of regenerating islet-derived proteins in inflammatory bowel disease. World J. Gastroenterol. 26, 2702–2714 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 49.Cao S., Su X., Zeng B., Yan H., Huang Y., Wang E., Yun H., Zhang Y., Liu F., Li W., Wei H., Che Y., Yang R., The gut epithelial receptor LRRC19 promotes the recruitment of immune cells and gut inflammation. Cell Rep. 14, 695–707 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 50.Quigley E. M. M., Microbiota-brain-gut axis and neurodegenerative diseases. Curr. Neurol. Neurosci. Rep. 17, 94 (2017). [DOI] [PubMed] [Google Scholar]
  • 51.Seo D. O., O'Donnell D., Jain N., Ulrich J. D., Herz J., Li Y., Lemieux M., Cheng J., Hu H., Serrano J. R., Bao X., Franke E., Karlsson M., Meier M., Deng S., Desai C., Dodiya H., Lelwala-Guruge J., Handley S. A., Kipnis J., Sisodia S. S., Gordon J. I., Holtzman D. M., ApoE isoform- and microbiota-dependent progression of neurodegeneration in a mouse model of tauopathy. Science 379, eadd1236 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 52.Ferreiro A. L., Choi J., Ryou J., Newcomer E. P., Thompson R., Bollinger R. M., Hall-Moore C., Ndao I. M., Sax L., Benzinger T. L. S., Stark S. L., Holtzman D. M., Fagan A. M., Schindler S. E., Cruchaga C., Butt O. H., Morris J. C., Tarr P. I., Ances B. M., Dantas G., Gut microbiome composition may be an indicator of preclinical Alzheimer's disease. Sci. Transl. Med. 15, eabo2984 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 53.Paneni F., Beckman J. A., Creager M. A., Cosentino F., Diabetes and vascular disease: Pathophysiology, clinical consequences, and medical therapy: Part I. Eur. Heart J. 34, 2436–2443 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 54.Nestel P., Goldrick B., Obesity: Changes in lipid metabolism and the role of insulin. Clin. Endocrinol. Metab. 5, 313–335 (1976). [DOI] [PubMed] [Google Scholar]
  • 55.Del Campo M., Peeters C. F. W., Johnson E. C. B., Vermunt L., Hok A. H. Y. S., van Nee M., Chen-Plotkin A., Irwin D. J., Hu W. T., Lah J. J., Seyfried N. T., Dammer E. B., Herradon G., Meeter L. H., van Swieten J., Alcolea D., Lleo A., Levey A. I., Lemstra A. W., Pijnenburg Y. A. L., Visser P. J., Tijms B. M., van der Flier W. M., Teunissen C. E., CSF proteome profiling across the Alzheimer's disease spectrum reflects the multifactorial nature of the disease and identifies specific biomarker panels. Nat Aging 2, 1040–1053 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 56.van der Ende E. L., In ‘t Veld S. G. J. G., Hanskamp I., van der Lee S., Dijkstra J. I. R., Hok-A-Hin Y. S., Blujdea E. R., van Swieten J. C., Irwin D. J., Chen-Plotkin A., Hu W. T., Lemstra A. W., Pijnenburg Y. A. L., van der Flier W. M., Campo M. D., Teunissen C. E., Vermunt L., CSF proteomics in autosomal dominant Alzheimer's disease highlights parallels with sporadic disease. Brain 146, 4495–4507 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 57.Ewers M., Franzmeier N., Suarez-Calvet M., Morenas-Rodriguez E., Caballero M. A. A., Kleinberger G., Piccio L., Cruchaga C., Deming Y., Dichgans M., Trojanowski J. Q., Shaw L. M., Weiner M. W., Haass C., I. Alzheimer's Disease Neuroimaging, Increased soluble TREM2 in cerebrospinal fluid is associated with reduced cognitive and clinical decline in Alzheimer's disease. Sci. Transl. Med. 11, eaav6221 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 58.Zhao A., Jiao Y., Ye G., Kang W., Tan L., Li Y., Deng Y., Liu J., Alzheimer’s I.; Alzheimer’s Disease Neuroimaging Initiative (ADNI) , Soluble TREM2 levels associate with conversion from mild cognitive impairment to Alzheimer's disease. J. Clin. Invest. 132, e158708 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 59.Chen J., Doyle M. F., Fang Y., Mez J., Crane P. K., Scollard P., Core A. D. H. C. C. H., Satizabal C. L., Alosco M. L., Qiu W. Q., Murabito J. M., Lunetta K. L., Peripheral inflammatory biomarkers are associated with cognitive function and dementia: Framingham heart study offspring cohort. Aging Cell 22, e13955 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 60.Wik L., Nordberg N., Broberg J., Björkesten J., Assarsson E., Henriksson S., Grundberg I., Pettersson E., Westerberg C., Liljeroth E., Falck A., Lundberg M., Proximity extension assay in combination with next-generation sequencing for high-throughput proteome-wide analysis. Mol. Cell. Proteomics 20, 100168 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 61.Henson R. L., Doran E., Christian B. T., Handen B. L., Klunk W. E., Lai F., Lee J. H., Rosas H. D., Schupf N., Zaman S. H., Lott I. T., Fagan A. M., Cerebrospinal fluid biomarkers of Alzheimer's disease in a cohort of adults with Down syndrome. Alzheimers Dement. (Amst) 12, e12057 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 62.Cruchaga C., Western D., Timsina J., Wang L., Wang C., Yang C., Ali M., Beric A., Gorijala P., Kohlfeld P., Budde J., Levey A., Morris J., Perrin R., Ruiz A., Marquie M., Boada M., de Rojas I., Rutledge J., Oh H., Wilson E., Guen Y. L., Alvarez I., Aguilar M., Greicius M., Pastor P., Pulford D., Ibanez L., Wyss-Coray T., Sung Y. J., Phillips B., Proteogenomic analysis of human cerebrospinal fluid identifies neurologically relevant regulation and informs causal proteins for Alzheimer's disease. Res. Sq, rs.3.rs-2814616 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Tables S1 to S9

Legends for data S1 to S8

sciadv.adk3674_sm.pdf (423.4KB, pdf)

Data S1 to S8

sciadv.adk3674_data_s1_to_s8.zip (2.9MB, zip)

Articles from Science Advances are provided here courtesy of American Association for the Advancement of Science

RESOURCES