Back

Predicting Alzheimer's disease phenotypes with aging clocks: an exploratory analysis

David Sarmento, C.; Drouard, G.; Saari, T. T.; Aaltonen, A.; Heikkinen, A.; Palviainen, T.; Herukka, S.-K.; Kokkola, T.; Kärkkäinen, S.; FinnGen, ; Palotie, A.; Julkunen, V.; Runz, H.; Kaprio, J.; Ollikainen, M.; Vuoksimaa, E.

2025-10-22 genetic and genomic medicine
10.1101/2025.09.01.25334848 medRxiv
Show abstract

BackgroundThe continuous aging of the worlds population urges the improvement of early diagnosis of age-related diseases, such as Alzheimers disease (AD). Blood proteomes can detect systemic and organ-specific disease-related changes and are accessible by minimally invasive blood draws. MethodsWe explored the potential of proteomic and epigenetic aging clocks as complementary biomarkers to established AD-related blood-based biomarkers (BBBs) and cognitive tests. Omics were generated from blood samples of 153 cognitively unimpaired individuals (average age 62 years). We investigated the associations of biological aging with BBBs and cognition, measured approximately nine years after the omics. We additionally tested whether dementia risk factors or genetic liability to them modulated these associations. FindingsAccelerated systemic and brain-specific proteomic aging were linked with poorer cognitive functioning and higher plasma levels of neurofilament light chain. Interaction analysis showed that negative associations between proteomic aging and cognitive scores were stronger in individuals with lower genetic liability for type II diabetes. Altogether, proteomic clocks improved the explained variance in cognitive and biomarker measures by up to 18% compared with epigenetic clocks alone. InterpretationOur results support the potential of proteomes in detecting aging and AD-related phenotypes, particularly neurodegeneration and cognitive decline. However, co-morbidities may constitute confounding factors, highlighting the importance of further investigating proteomic aging in the context of AD using comprehensive approaches. FundingFIMM-EMBL International PhD Programme; Sigrid Juselius Foundation; Academy of Finland; FinnGen.

Matching journals

The top 5 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.