Late-onset Alzheimer's disease (LOAD) is a common irreversible neurodegenerative disease with heterogeneous genetic characteristics. Identifying the biological biomarkers with the potential to predict the conversion from normal controls to LOAD is clinically important for early interventions of LOAD and clinical treatment. The polygenic risk score for LOAD (AD-PRS) has been reported the potential possibility for reliably identifying individuals with risk of developing LOAD recently. To investigate the external phenotype changes resulting from LOAD and the underlying etiology, we summarize the comprehensive associations of AD-PRS with multiple biomarkers, including neuroimaging, cerebrospinal fluid and plasma biomarkers, cardiovascular risk factors, cognitive behavior, and mental health. This systematic review helps improve the understanding of the biomarkers with potential predictive value for LOAD and further optimizing the prediction and accurate treatment of LOAD.