In a recent study published in the journal npj Aging, researchers developed a urinary microRNA (miRNA) aging clock that predicts chronological age from urinary miRNA profiles, with deviations used as an indirect indicator of biological age acceleration rather than a direct measurement of biological age.

Aging remains the primary driver of chronic diseases, yet reliable, non-invasive biomarkers of biological age are limited. Aging clocks are biomarker-based models that estimate biological age from age-responsive characteristics, and deviations from chronological age reflect the pace of aging. These clocks are associated with morbidity and mortality risk and are used to stratify health risk and assess interventions.

miRNAs have been linked to age-related disorders and aging. Aging clocks based on miRNAs in skin, blood, and plasma have been developed. Beyond skin, solid tissues exhibit reproducible age-related miRNA shifts. While tissue or blood sampling requires invasive procedures at medical facilities, urine represents a non-invasive, scalable, and inexpensive source of extracellular vesicle (EV) miRNAs, yet it remains underexplored and is subject to pre-analytical variability related to hydration, urine concentration, and collection conditions.

Study Design and Population Sampling
In the present study, researchers developed and validated an aging clock based on urinary EV miRNA (uEV-miRNA). Urine samples were collected from 6,331 Japanese individuals undergoing a cancer screening test in a general population setting rather than a clinically adjudicated disease cohort. Data on age, sex, height, weight, smoking and alcohol status, exercise, and comorbidities were collected using a questionnaire. Comorbidities were self-reported, and urine samples were processed to remove cells and debris and extract EVs. RNA was extracted, and small RNA libraries were prepared.

Sequencing and Bioinformatic Processing
Library concentrations were quantified, and equimolar libraries were sequenced. Reads were sequentially aligned to human small RNA references in the following order: miRNA, transfer RNA (tRNA), tRNA-derived fragments (tRFs), piwi-interacting RNA (piRNA), and ribosomal RNA (rRNA). Reads not mapping to any of these small RNA references were aligned to human messenger RNAs (mRNAs) and the hg19 genome.

A filtering threshold of at least 100,000 total miRNA counts per sample was applied. In addition, analyses were restricted to miRNAs present in at least half of all samples to exclude sporadically expressed or rare miRNAs. The team created training and test datasets using participants’ metadata and propensity score matching. The training dataset included 2,400 participants, and the test dataset included 2,840 matched individuals (test set 1).

Model Development and Validation Strategy
An independent dataset, test set 2, of 1,091 individuals was assembled for validation. Age was modeled from urinary miRNA expression using a light gradient boosting machine (LightGBM) regressor. Following training and internal validation, each model was evaluated on test sets 1 and 2. Raw predictions from the test sets were corrected to generate urinary miRNA ages. Sex-specific mean absolute errors (MAEs) were estimated against chronological age.

The difference between miRNA age and chronological age was defined as biological age acceleration (ΔAge). The top 20 miRNAs were ranked based on their LightGBM feature importance. Age-related trends in these top-ranked miRNAs were evaluated. The PubMed database was queried to assess representation of the top-ranked miRNAs in the aging literature. Finally, the team investigated whether ΔAge is associated with common comorbidities, noting that estimates in individuals younger than 25 years or older than 80 years require cautious interpretation.

Aging Clock Performance and Accuracy
uEV-miRNA reads represented the largest fraction in urine samples, 50% to 60%, and contributions from rRNA, tRNA, piRNA, tRFs, and mRNA/hg19 reads were relatively consistent. Following exclusion of sporadically expressed or rare uEV-miRNAs, 407 miRNA features were included in clock development. The uEV-miRNA aging clock achieved an MAE of 5.1 years on the training set, 4.5 years on test set 1, and 4.4 years on test set 2.

Key miRNAs and Biological Pathways
Mean expression levels across the top 20 miRNAs were clearly associated with age. Six miRNAs decreased with age, 10 increased with age, and four showed increases specifically in males. Most of these top miRNAs were significantly enriched in cellular senescence and aging. Notably, this set included well-established aging-associated miRNAs, geromiRs, such as miR-155-5p, miR-146a-5p, miR-31-5p, and miR-34a-5p.

A gene ontology (GO) analysis indicated significant enrichment of pathways related to cellular senescence and aging, with notably enriched GO terms including bone remodeling, regulation of osteoclast development, and marginal zone B cell differentiation. Among comorbidities assessed for associations with ΔAge, only type 2 diabetes showed a significant increase. Specifically, ΔAge was elevated in females aged 50 to 69 and males aged 50 to 79 years. The authors note that other comorbidities may not have shown associations because conditions were reported as medical history and may not have been actively present at the time of urine collection.

Interpretation, Limitations, and Clinical Relevance
Overall, the uEV-miRNA aging clock showed an MAE of approximately 4.4-5.1 years across datasets. Top-ranked uEV-miRNAs included canonical geromiRs, which are consistently upregulated in aged tissues and senescent cells. Although less accurate than DNA methylation-based clocks, the urinary miRNA aging clock matched or exceeded the performance of several previously reported blood-based miRNA and mRNA clocks in larger cohorts.

The authors caution that ΔAge should be interpreted as a general risk indicator rather than a disease-specific or diagnostic measure, particularly in individuals with active malignancy or urogenital pathology. Disease-specific validation and recalibration will be required before clinical use. Raw sequencing data are not publicly available, and further independent replication will be important for broader translational adoption. These findings establish uEV-miRNAs as robust, scalable, and non-invasive biomarkers of biological age.

Journal reference:
Havelka, M., Satomura, A., Yamaguchi, H., et al. (2025). A urinary microRNA aging clock accurately predicts biological age. npj Aging. DOI: 10.1038/s41514-025-00311-3, https://www.nature.com/articles/s41514-025-00311-3