Select Small Non‐Coding <scp>RNAs</scp> Are Determinants of Survival in Older Adults

Virginia Byers Kraus, Sisi Ma, Syeda Iffat Naz, Xin Zhang, Christopher G Vann, Nenad Bursac, W B. Kraus, Steven S. Shen, J. Huebner, Ching-Heng Chou, Erich Kummerfeld, H. J. Cohen, Constantin AliferisAging 2026 · DOI

TL;DR

This study identifies circulating small non-coding RNAs (smRNAs), particularly piwi-interacting RNAs (piRNAs), as key determinants of survival in older adults. The researchers developed highly accurate predictive models for 2-year survival using these smRNAs and identified specific piRNAs as potential therapeutic targets for promoting longevity.

What Problem Does This Paper Address?

Understanding the molecular underpinnings of human longevity and identifying reliable biomarkers for predicting survival in older adults is a critical challenge. While microRNAs (miRNAs) have been studied in this context, the role of piwi-interacting RNAs (piRNAs) in human aging and survival remains largely unexplored. This paper aims to bridge this gap by investigating the relevance of both miRNAs and piRNAs as epigenetic factors influencing longevity, developing predictive models for survival, and identifying potential therapeutic targets.

Methods

The study employed a multi-tiered validation approach, integrating predictive and causal modeling techniques. Key methods include:

  • Next-Generation Sequencing: Used to quantify the levels of 828 small non-coding RNAs (687 miRNAs and 141 piRNAs) in plasma samples.
  • Markov Boundary (MB) Analysis: A causal inference technique used to identify variables that are direct causes of survival. This method, rooted in causal graph theory, helps distinguish predictive signals from causal relationships.
  • Nested Cross-Validation (NCV): Employed to ensure robust model selection and unbiased error estimation, preventing overfitting.
  • External Validation: Crucial for assessing the generalizability of the developed models using an independent dataset sequenced in a separate batch.
  • Sepset Analysis: Used to understand why certain variables might not be selected in MBs and to clarify the information content of clinical variables versus smRNAs.
  • Target Prediction and Pathway Analysis: Utilized to identify potential mRNA targets of the identified smRNAs and to explore the biological pathways they influence.

Datasets

  • Duke-EPESE Cohort: The primary dataset comprised 1271 community-dwelling older adults (≥ 71 years) from the Duke-Established Populations for Epidemiologic Studies of the Elderly (D-EPESE) study.
  • Expanded Discovery Dataset (n=707): A subset of the D-EPESE cohort used for initial model building and internal validation. This was further split into:
  • Discovery Subset (n=505): Used for model construction.
  • Internal Validation Dataset (n=202): Used for initial validation within the same sequencing batch.
  • Independent External Validation Dataset (n=564): A separate half of the D-EPESE cohort, sequenced in a different batch, used to rigorously test the generalizability and robustness of the predictive models.

Key Insights and Ideas

  • PiRNAs as Key Longevity Determinants: The study strongly implicates piRNAs, not just miRNAs, as significant players in human survival, particularly in older adults. This challenges the traditional view of piRNAs primarily being involved in germline defense.
  • Reduced PiRNA Levels Correlate with Longer Lifespan: A surprising finding was that lower levels of certain piRNAs were associated with longer survival, suggesting a novel mechanism for longevity beyond their known role in transposon silencing.
  • Causal Evidence for smRNAs in Survival: The application of causal inference methods provides strong evidence that circulating smRNAs are not just correlates but potential direct causes of survival outcomes.
  • High Predictive Power for Short-Term Survival: The developed models, especially those incorporating smRNAs, demonstrate remarkable accuracy in predicting 2-year survival, outperforming models based on age and clinical variables alone.
  • Potential for Druggable Targets: The identification of specific piRNAs as potential therapeutic targets opens avenues for developing interventions to promote healthy aging and extend lifespan.

Key Results

  • Predictive Models:
    • A combined model of smRNAs, age, and clinical variables achieved a cross-validated AUC of 0.92 for 2-year survival in the discovery set and 0.87 in external validation.
  • Streamlined models using only six piRNAs achieved an external validation AUC of 0.83 for 2-year survival, demonstrating high predictive power with fewer features.
  • Predictive performance declined for 5- and 10-year survival horizons, highlighting the greater influence of smRNAs on short- to medium-term survival.
  • Causal Discovery:
    • Nine piRNAs were identified as potential direct causes of 2-year survival.
  • Four piRNAs, along with IADL-motor function and HDL particle concentration, consistently appeared in all MBs for 2-year survival, suggesting they are critical causal factors.
  • Seven smRNAs (2 miRNAs, 5 piRNAs) were identified as potential direct causes of 5-year survival.
  • Intervention Effects:
    • A hypothetical "intervention" to adjust smRNA levels to promote survival could increase the probability of 2-year survival from 47% to 90% (discovery) and 73% to 94% (external validation).
  • For 10-year survival, a similar intervention could increase survival probability from 25% to 77%.

Limitations and Future Work

  • Experimental Validation: While the study provides strong causal inference, the authors acknowledge that these findings do not replace experimental validation of the identified smRNAs and their targets.
  • PiRNA Isoforms: The detection of numerous piRNA isoforms (isopiRs) and the evolving understanding of their functional equivalence present a challenge.
  • PiRNA Target Prediction: Target prediction tools for piRNAs are still under development, requiring further refinement and validation.
  • Generalizability: While external validation was performed on a separate batch from the same cohort, validation in broader and more diverse cohorts is recommended.
  • 10-Year Survival Predictivity: The limited predictive power for 10-year survival suggests that other factors become more dominant over longer time horizons, or that current molecular measures are less informative for very long-term outcomes.
  • Future Work: The authors suggest exploring the relationship between smRNAs and DNA methylation patterns to bridge smRNA function with existing methylation-based aging clocks, and to investigate the identified piRNAs and miRNAs in mechanistic studies to elucidate their roles in longevity.

Summary updated

Abstract

To investigate the relevance of small RNAs to human longevity, we pursued three goals: (a) to validate epigenetic (small RNA) factors underlying survival of older adults, (b) to develop and validate prediction models of survival for potential clinical application, and (c) to identify plausible druggable targets prolonging longevity. We evaluated 828 small non-coding RNAs-687 microRNAs (miRNAs) and 141 piwi-interacting RNAs (piRNAs)-in baseline plasma from 1271 community-dwelling older adults (≥ 71 years) in the Duke-EPESE study. Our predictive model incorporating smRNAs, clinical variables (demographics, lifestyle, mood, physical function, standard clinical laboratory tests, NMR-derived lipids and metabolites, and medical conditions) and age achieved strong performance, with cross-validated AUCs of 0.92 for 2-year survival in Discovery and 0.87 in external Validation. Nine piRNAs, all reduced in longer-lived individuals, were identified as potential therapeutic targets. Under the assumption of causal sufficiency, these data provide causal evidence linking circulating small RNAs with survival outcomes in humans. While such inference does not replace experimental validation, it complements mechanistic studies by identifying candidate molecular drivers most relevant to human longevity. Supporting biological plausibility, reduced piRNA biogenesis has been shown to double lifespan in C elegans. Together, our findings identify circulating piRNAs and miRNAs as promising biomarkers and potential therapeutic targets to advance human longevity.
BibTeX
@inproceedings{kraus2026select,
  author = {Virginia Byers Kraus and Sisi Ma and Syeda Iffat Naz and Xin Zhang and Christopher G Vann and Nenad Bursac and W B. Kraus and Steven S. Shen and J. Huebner and Ching-Heng Chou and Erich Kummerfeld and H. J. Cohen and Constantin Aliferis},
  title = {{Select Small Non‐Coding <scp>RNAs</scp> Are Determinants of Survival in Older Adults}},
  booktitle = {Aging},
  year = {2026},
  doi = {10.1111/acel.70403},
}
Figure