Individual circulating microRNAs and continuous cognitive performance in a mixed inpatient-outpatient geriatric...

Circulating microRNAs have been proposed as minimally invasive markers of cognitive impairment and neurodegenerative disease, but evidence for individual miRNAs in heterogeneous older populations remains limited.

We examined cross-sectional associations between eight circulating serum miRNAs and continuous cognitive performance in a mixed inpatient-outpatient geriatric cohort.

This cross-sectional analysis included 65 enrolled participants from a single-center, non-consecutive, mixed inpatient-outpatient geriatric cohort; 63 formed the analytic cohort (32 inpatients, 31 outpatients). Eligibility required written informed consent, availability of a routine-care Mini-Mental State Examination (MMSE) score, administration of Addenbrooke's Cognitive Examination III (ACE-III), fasting serum sampling, and structural brain MRI during the same visit or within 48 hours. Eight predefined serum miRNAs were quantified by RT-qPCR with exogenous spike-in controls, with cel-miR-39-3p used as the normalization reference. The primary outcome was continuous MMSE score. Secondary outcomes were ACE-III total score, the ACE-III attention/orientation subscale, and threshold-defined MMSE and ACE-III impairment categories. Multivariable linear or logistic regression models were adjusted for age, sex, years of education, and recruitment setting. The main analysis used adjusted linear regression for continuous MMSE with Benjamini-Hochberg false discovery rate correction across the eight individual-miRNA MMSE models. Bootstrap was used to assess the stability of the MMSE estimates. Sensitivity analyses for the three strongest MMSE associations included Huber robust regression and miRNA by setting interaction testing.

In multivariable models adjusted for age, sex, years of education, and recruitment setting, higher levels of miR-132, miR-545, and miR-34a were linked to lower MMSE scores. A 1-SD increase in each marker corresponded to an MMSE decrease of about 1.1-1.3 points. These were the strongest single-miRNA findings in the dataset and met the prespecified FDR threshold in the primary models, although bootstrap analyses showed limited precision. Robust regression gave the same negative pattern for all three markers, suggesting that the results were not driven by a small number of extreme observations. Evidence of between-setting differences was seen only for miR-132 and only at the nominal level. No individual miRNA showed a clear nominal association with ACE-III total, and the ACE-III attention/orientation and MRI-adjusted analyses were exploratory. In secondary threshold-based models, no single miRNA remained significant after FDR correction.

In this cross-sectional pilot cohort, higher miR-132, miR-545, and miR-34a were associated with lower MMSE scores in adjusted analyses. However, effect sizes were small, bootstrap analysis showed considerable uncertainty, and the findings are exploratory and hypothesis-generating. These results require evaluation in larger consecutive cohorts and do not establish diagnostic, prognostic, or standalone clinical biomarker utility.