Pondering a Paper: Organ Aging Signatures
New Research on Organ-Specific Aging Biomarkers
Paper: Organ aging signatures in the plasma proteome track health and disease
How do people age differently, and how best should we track aging for future clinical applications? These are two questions that after decades of research the field hasn’t come to a consensus on. A large focus has been on trying to find whole-body biomarkers of aging, instead of organ-specific biomarkers. One of the main conclusions of an impactful paper that came out of the Wyss Coray lab at Stanford last year, Organ aging signatures in the plasma proteome track health and disease, is that looking at signs of aging in each organ individually matters. This work introduced a multi-cohort dataset with measurements of ~ 5,000 bloodborne proteins, plus follow-up analyses to discover aging biomarkers. A large number of aging biomarkers they found were organ-specific. Using these biomarkers, they found that different organs can age at different rates, sometimes dramatically.
I asked Jarod Rutledge, one of the co-first authors of the paper, about the significance of his work. He said “Aging is not a monolith, it happens in different ways for different people. Human longevity-focused medicine will need to be personalized for individual needs, and to be useful in a clinical or preventative health setting biomarkers of aging need to reflect this. The ability to accurately measure aging of specific sub-systems in the body and predict future function is a crucial step to clinical use. Our work lays the foundation for this by showing that it is possible to develop interpretable aging biomarkers for specific organs and subsystems, and that doing so can give us more powerful predictions about future health and disease. We believe that extensions of this approach will be used to guide clinical trials and routine care in the future.”
Introduction to this paper and methods
The authors of the paper identified a number of bloodborne proteins that predict biological age at the individual-organ level, created predictive models using these proteins, and tried to demonstrate the association between accelerated aging within an organ and future disease. I’ll discuss a bit about the particular methods they used next.
They measured levels of ~ 5,000 bloodborne proteins using samples from five aging and Alzheimer's disease (AD) cohorts. Quantification was performed using a commercially available test - the Somalogic Somascan Assay. This assay, described more in previously published work and on the Somalogic website, uses chemically modified short DNA strands that selectively bind individual proteins. These strands are attached to a fluorescent chemical group, and the concentration of protein relates to the amount of observed fluorescence.
To identify which proteins were
correlated with aging in a particular organ, the authors first analyzed published research from bulk RNA-sequencing experiments. They identified proteins with RNA transcripts that showed at least a 4x log-fold change of expression in a specific organ compared to other organs, and ~850 proteins satisfied this criterion.
For each organ, they trained predictive models to estimate the participant's age just using that organ’s specific proteins. They also trained two other non-organ-specific predictive models to predict biological age: one using all proteins not specific to an organ (referred in the paper as the "organismal aging" model) and one using all proteins in the dataset (the "conventional" model). They applied these models to four datasets. Knowing the actual age of each participant, they calculated an "age gap" for each organ and for general aging by subtracting the predicted age from the actual age. For each individual and each organ, participants were classified as "extreme aging types" if their age gap was greater than 2 standard deviations from the mean. They found that about one-fifth of people were extreme agers for at least one organ, while only approximately 2% had multiple organs with extreme aging signatures.
The next step in their analysis was to relate organ aging to specific age-related diseases. Some of the cohorts they collected blood proteomic measurements on included longitudinal health data collected over multiple years, which let them relate the biomarker data collected at one time to the risk of developing future diseases years later. These tests helped validate that they were measuring organ-specific signatures of future disease.
My thoughts
I thought this paper was well-written, and besides its scientific findings, it was interesting because it is easy to see how it could be translated to clinical care. The samples they used in the study were collected noninvasively - they were just blood samples. You could imagine taking a blood test that quantifies your levels of organ-specific aging proteins, and then getting a read-out on what that means in terms of your different organ ages overall. If any organs show signs of accelerated aging, you or your doctor could then order more intensive organ-specific tests.
However, I had some thoughts about things that could have been done differently/unanswered questions:
Some of the biomarkers they claim are aging signatures could be pre-symptomatic signatures of common diseases, and the paper doesn’t address this issue. Functional validation is needed to actually uncover how identified proteins affect physiology. If associated with a common disease, an identified protein would be useful as a diagnostic but not be directly related to aging itself, since a protein that is associated with aging would lead to rates of all diseases of an organ system increasing as opposed to rates of a particular disease only increasing.
The cohort they used to train their models (Knight ADRC healthy controls) may be too old to uncover some biomarkers of aging before they start causing deleterious effects in the body, regardless of whether they result in functional symptoms. The mean age of the cohort is 73, with a standard deviation of approximately 10 years (Supplementary Table 1 of the paper). This age range might miss early aging biomarkers that could be present in younger populations.
The number of proteins is smaller than current state-of-the-art methods allow. Somalogic currently offers 11k proteins in their assays (compared to the 5k used in the paper). I would be interested in how increasing the number of proteins impacts both the predictive performance of biological aging models, and the number of additional biomarkers that would have been found. The authors acknowledge this limitation in the paper’s discussion section.
To identify organ-specific proteins, it would have been better to use proteomics as opposed to transcriptome measurements. This is because measurements of RNA-seq are indirectly correlated to protein measures.
References/Sources Used
Oh, H.SH., Rutledge, J., Nachun, D. et al. Organ aging signatures in the plasma proteome track health and disease. Nature 624, 164–172 (2023). https://doi.org/10.1038/s41586-023-06802-1
The GTEx Consortium ,The GTEx Consortium atlas of genetic regulatory effects across human tissues. Science. 369,1318-1330(2020).DOI:10.1126/science.aaz1776
Kraemer, Stephan, et al. "From SOMAmer-based biomarker discovery to diagnostic and clinical applications: a SOMAmer-based, streamlined multiplex proteomic assay." PloS one 6.10 (2011): e26332.
Vogel, C., Marcotte, E. Insights into the regulation of protein abundance from proteomic and transcriptomic analyses. Nat Rev Genet 13, 227–232 (2012). https://doi.org/10.1038/nrg3185
 | A guest post by
|