Plasma proteomic signatures of preclinical Alzheimer's disease biomarkers and memory in clinically unimpaired older adults.

Background: Multianalyte plasma proteomic panels that can accurately detect initial AD pathology in preclinical populations and simultaneously measure related biological processes relevant for disease risk are critical for advancing early detection and prognosis. Methods: Using the NULISAseq CNS panel, we measured plasma from 193 clinically unimpaired (CU) older adults enrolled in the Stanford Aging and Memory Study (SAMS). We evaluated correspondence of core AD-relevant biomarkers Aβ42, Aβ40, pTau217, pTau181, GFAP, NfL, Aβ42/Aβ40, and pTau217/Aβ42 measured using NULISAseq and established Lumipulse immunoassays. ROC curve analyses compared the accuracy of these biomarkers for detecting Lumipulse CSF Aβ-positivity across platforms. Linear models were applied across 124 NULISAseq proteins to examine associations with common AD risk factors, including age, female sex, and APOE-ε4, as well as with biomarkers CSF Aβ42/Aβ40 and pTau181, 18F-PI2620 Tau PET, and memory. Fold change differences in NULISAseq proteins as a function of CSF Aβ (A+) and pTau181 (T+) status were examined using Wilcoxon rank-sum tests. Results: Moderate to high correlations were observed between NULISAseq and Lumipulse AD plasma biomarkers. Across platforms, plasma pTau217/Aβ42 exhibited the highest performance in discriminating CSF A+ (NULISAseq AUC: 0.940, 95%CI: 0.885-0.995; Lumipulse AUC: 0.907, 95%CI: 0.849-0.966). Age and sex were associated with differential expression of NULISAseq targets linked to neurodegeneration, microglial activation, and inflammation. CSF A+ was associated with fold change differences in Aβ42, pTau217, pTau231, pTau181, and GFAP, while CSF T+ was additionally associated with increases in TREM1, TIMP3, SAA1, and S100A12. When stratified by AT groups, A+T- exhibited lower Aβ42 and elevated pTau217 compared to A-T-, whereas A+T+ exhibited elevated pTau231 and pTau181 compared to A+T-. Temporal cortex tau was positively associated with NULISAseq pTau217, pTau231, pTau181, and pTau217/Aβ42. Memory function was negatively associated with pTau isoforms and PRDX6, YWHAZ, ENO2, ARSA, CHI3L1, CXCL8, and FCN2. These associations remained when controlling for pTau217 and restricting to A- CU, suggesting these targets may represent AD-independent biological pathways relevant for memory function. Conclusions: NULISAseq immunoassay-based multiplexing accurately detects AD pathology among CU older adults and identifies multiple biological pathways related to early biomarker abnormality and memory function that may become dysregulated in preclinical AD.