Long-term temporal stability of circulating proteins in older adults

Abstract

Circulating proteins reflect diverse biological processes and can offer critical insights into an individual's overall health and aging trajectory. The circulating proteome is shaped by a complex interplay of genetic, biological, and environmental factors across the lifespan. However, little is known about which factors influence its long-term temporal stability. Here we used SomaScan proteomics to evaluate the five-year temporal stability of 7,288 proteins measured in serum from 3,093 participants (mean age 76 years) of the Age, Gene/Environment Susceptibility (AGES)-Reykjavik study. We observed a wide variability in the temporal stability of individual proteins, with temporally stable proteins more often being extracellular and associated with diseases, while temporally variable proteins are typically involved in intracellular housekeeping functions. We demonstrate that temporal stability of circulating proteins does not reflect that of transcriptomic stability in tissues, and that genetic effects and disease stage are two major contributors to protein temporal stability in the circulation. Our findings underscore the protein-specific differences in long-term temporal stability, and the genetic and biological factors influencing them, which are particularly important to consider in the context of biomarker development and precision medicine.

Figure 1:. Temporal stability of proteins (n = 7,288).

The temporal stability of proteins was assessed using temporal SCC estimates. Proteins were divided into three groups: temporally variable (<0.40), medium (0.40–0.75), and stable (>0.75). A) Distribution of temporal SCC estimates of all proteins. B – D) Examples of proteins from each group, showing baseline (AGES) versus follow-up (AGESII) protein levels for each individual. B) LILRA6 was the most temporally stable protein (temporal SCC = 0.97). C) PDYN showed medium temporal stability (temporal SCC = 0.57). D) APOH was temporally variable (temporal SCC = 0.14).

Figure 2:. Enrichment and functional analysis using g:Profiler.

A) Comparison of enriched pathways between temporally stable and variable protein groups, showing distinct biological roles with a few pathways shared across groups. Data points are colored based on whether temporally stable, variable proteins, or both significantly enriched the pathway. B) Top 15 enriched Gene Ontology (GO) driver terms for temporally stable proteins. C) Enriched pathways for temporally stable proteins from KEGG, Reactome, and WikiPathways. D) Top 15 GO driver terms for temporally variable proteins. E) Enriched pathways for temporally variable proteins from KEGG, Reactome, and WikiPathways.

Figure 3:. Functional characteristics of temporally stable and variable proteins.

A) Protein classes and tissue-enriched gene/protein expression (x-axis log-scaled). Odds ratio estimates are presented with 95% confidence intervals. B) Loss-of-function (LoF) intolerance, with variable proteins showing significantly higher LoF intolerance (Wilcoxon test, two-sided, p < 1 × 10³). C) Degree (log-scaled) in the protein–protein interaction (PPI) network, where variable proteins had significantly higher connectivity, indicating they are more likely to be hubs. D) Degree (log-scaled) in the co-regulatory serum network, where variable proteins also showed higher connectivity (Wilcoxon test, two-sided, p < 1 × 10³). Boxplots in (B–C) indicate median value, 25^th and 75^th percentiles. Whiskers extend to smallest/largest value of no further than 1.5 & interquartile range. Outliers are not shown.

Figure 4:. Effect of strong pQTLs on temporal stability.

A) Scatter plot of temporal SCC estimates before and after adjusting protein levels for conditionally independent pQTLs (p < 5 × 10). Original and adjusted SCC values were strongly correlated (SCC = 0.94), but 75% of SOMAmers showed lower SCC values after adjustment. Proteins are colored by pQTL type, with less significant points shown at reduced opacity. B–C) Examples of proteins initially classified as temporally stable. B) SPINT3 remained temporally stable after adjustment (original SCC = 0.84; adjusted SCC = 0.85). C) ACP1 (original SCC = 0.85) initially showed genotype-based separation before adjustment but was reclassified as temporally variable after adjustment (SCC = 0.35), with reduced genotype separation and more apparent within-individual fluctuation.

Figure 5:. Baseline protein module enrichment for temporal stability protein groups.

A) Examples of modules enriched for each temporal stability (TS) group. Proteins in module M16 had significantly higher median temporal SCC estimates compared to other proteins (Wilcoxon test, two-sided, p < 1 × 10³). In contrast, proteins in module M8 had lower median temporal SCC estimates compared to other proteins (Wilcoxon test, two-sided, p < 1 × 10³). Boxplots indicate median value, 25^th and 75^th percentiles. Whiskers extend to smallest/largest value of no further than 1.5 & interquartile range. Outliers are not shown. B) Sankey plot showing how proteins are grouped into modules at each time point. Some proteins remain in the same modules over time, while others shift between modules. Modules enriched for variable proteins are colored green, stable-enriched modules are colored blue, and modules without significant enrichment are colored grey.

Figure 6:. Temporal stability and phenotype associations.

A) Number of proteins measured at baseline that are significantly associated with each phenotype (excluding proteins with “medium” temporal stability), shown for both incident and prevalent disease outcomes. Bars are colored by protein temporal stability. An asterisk (*) above a phenotype indicates significant enrichment (FDR < 0.05) of either temporally stable or variable proteins associated with that phenotype. B) Modules labeled on the x-axis, ordered by median temporal SCC. The y-axis lists each phenotype tested for associations with baseline protein levels. Orange dots indicate significant enrichment for a phenotype within a module, with dot size reflecting the strength of enrichment (odds ratio). Modules are highlighted vertically, with green indicating temporally variable-enriched and blue indicating temporally stable-enriched. C) Disease score by temporal SCC enrichment of modules. The score was calculated as the number of associated phenotypes multiplied by the median odds ratio. No significant difference was observed between enrichment categories (Kruskal–Wallis p = 0.54), though some stable-enriched modules exhibited high scores. Boxplots indicate the median, 25th, and 75th percentiles; whiskers extend to the smallest and largest values within 1.5× the interquartile range. Outliers are not shown. All data points are displayed with jitter, and labeled points indicate top modules.

Figure 7:. Effect of temporal SCC on disease progression.

A–D) Three modules included protein groups associated with disease that differed in temporal stability depending on disease stage. For individuals diagnosed between visits (between A1, the baseline visit, and A2, the follow-up visit), there was a significant difference in median temporal SCC (empirical p < 0.01) compared to the other groups: those diagnosed before A1, those diagnosed after A2, or those who were never diagnosed. Two modules were temporally stable-enriched: M16 (A–B) and M2 (C), both of which showed reduced temporal SCC in the between-visits group. The temporally variable-enriched module M8 (D) showed the opposite pattern, with increased temporal SCC in individuals diagnosed between visits.