This week, many aging-research labs are running the same kind of “two-track” analysis: one track for epigenetic clocks, another for inflammatory readouts, then asking whether they actually meet in the middle. A new Cell Genomics study does exactly that, and it does so with a specific inflammatory axis—CXCL9 and TNF—that appears to behave less like a bystander and more like a lever.

Section 1: What the study actually tested in BCG-PRIME

The paper starts from a familiar premise in inflammaging research: chronic, low-grade inflammation often travels with aging-related pathology, but it can also show up alongside internal factors that are not directly tied to a specific disease diagnosis. To probe that overlap, the team analyzed how systemic inflammation connects to four epigenetic aging clocks, focusing on whether inflammatory biomarkers track with epigenetic aging acceleration.

The dataset at the center of the work is the BCG-PRIME cohort. The participants included in the main analysis had at least one comorbidity known to worsen length of life or quality of life. The study then examined frailty and multimorbidity in relation to epigenetic aging acceleration (EAA), using both “calendar-age-based” and “healthspan-based” clock designs.

The researchers used four epigenetic clocks. Horvath and Hannum are built around chronological age. PhenoAge is designed to align more closely with healthspan. GrimAge is a healthspan-based clock that strongly predicts all-cause mortality. The analysis also included acceleration versions of each clock, expressed as epigenetic aging acceleration (EAA) for each clock.

In the first pass, the team asked how frailty relates to EAA across the four clocks. Frailty showed connections with every clock except EAA_Hannum and EAA_Horvath. Even more striking, frailty appeared more strongly associated with GrimAge than with chronological-age-based measures. The same pattern held when the team moved from frailty to multimorbidity: GrimAge again produced the strongest associations.

Because smoking is a known confounder in both epigenetic damage and chronic inflammatory lung disease, the researchers adjusted for smoking. The GrimAge-linked results persisted after that adjustment, which matters because smoking can simultaneously influence epigenetic signatures and inflammatory pathways.

The paper then ties this to a specific disease context. Smoking is known to be involved in both epigenetic damage and COPD (chronic obstructive pulmonary disease). The study reports that COPD itself also showed a strong connection with GrimAge.

One-sentence conclusion for this section: In BCG-PRIME, GrimAge is the epigenetic clock that most consistently tracks frailty, multimorbidity, and COPD, even after smoking adjustment.

Section 2: So what is different about the inflammatory signals

If the first section tells you where the clocks “move,” the second section answers what inflammatory markers are actually attached to that movement. Here, the study uses 64 inflammation biomarkers as clues, but not all biomarkers are treated equally. The team restricted attention to items that passed quality-control checks and were deemed “measurable” in the analysis.

When the researchers compared these 64 inflammation biomarkers against the acceleration versions of the clocks, a split emerged.

EAA_Hannum and EAA_Horvath did not show significant connections with the inflammatory biomarker set. In contrast, the “general versions” of Horvath and Hannum—rather than their acceleration forms—were associated with multiple proteins. GrimAge and PhenoAge showed even stronger relationships.

The inflammatory proteins that repeatedly surfaced were not random. The paper highlights CCL family members, described as inflammatory cytokine–related proteins, as being observed alongside both GrimAge and PhenoAge signals and their EAA variants. HGF (hepatocyte growth factor) also appeared in the signal.

TNF (tumor necrosis factor) was another key observation. TNF showed up in the inflammatory pattern, and the study reports that TNF’s association was stronger for GrimAge and PhenoAge than for chronological age.

But the strongest relationship in the biomarker-to-clock mapping came from a single chemokine: CXCL9. The paper reports the strongest relationship between GrimAge and CXCL9, and it also notes that CXCL9 correlates with age acceleration.

One-sentence conclusion for this section: The inflammatory link that best explains GrimAge acceleration centers on CXCL9, with TNF and CCL-family proteins reinforcing the same direction.

Section 3: Mendelian randomization turns CXCL9 from correlation into causality

Correlation can be a trap in aging biology, especially when inflammation and epigenetic change can both be downstream of the same unmeasured factor. To address that, the team adds a causal-check layer using Mendelian randomization.

In the Mendelian randomization analysis, the researchers test whether genetically predicted differences in CXCL9 levels correspond to differences in epigenetic aging outcomes. The result is presented as a directionally consistent causal pattern: people with higher CXCL9 levels show epigenetically “older” profiles and faster aging.

Here, “faster” is interpreted through acceleration metrics such as EAA_GrimAge. The study reports that this causal connection is also confirmed for PhenoAge, not just GrimAge.

The paper further states that the association persists even after multiple statistical adjustments, which is important because multiple testing can inflate apparent effects when many comparisons are made.

The causal-check does not stop at CXCL9. The researchers also report that TNF and CXCL10 are causally linked to faster aging when the outcome is defined using EAA_GrimAge.

The study then uses these causal signals to support a broader claim about mechanism at the level of aging biology. Because inflammatory factors connect to multimorbidity, the authors conclude that inflammaging may “drive” age-related diseases rather than merely accompany them.

One-sentence conclusion for this section: Mendelian randomization supports CXCL9 as a causal driver of faster epigenetic aging, with TNF and CXCL10 showing similar causal links under GrimAge acceleration.

Section 4: Gender differences are small, and IFN/IL-22 responses fail to rise

The final section shifts from statistical inference to biological plausibility, and it does so by testing whether the inflammatory clock signals correspond to functional immune responses.

First, the team examines sex-stratified results. They report that men and women show little difference in reaching statistical significance. The CXCL9 effect appears to be largely independent of sex, suggesting that the CXCL9–aging link is not confined to one demographic group.

Next comes a functional experiment. The researchers use cells derived from the 500FG and BCG-PRIME cohorts and expose them to pathogens. The key observation is that cells that appear “older” according to the epigenetic clocks perform worse when challenged by pathogens.

This matters because it connects epigenetic aging signatures to immune competence. Even when overall inflammation is higher, the study reports that the actual immune response to threat does not scale appropriately.

Specifically, despite elevated total inflammation, IFN-γ and IL-22 responses fail to emerge properly in the “older” epigenetic clock context. The paper frames the work as observational for the clock–biomarker relationships and emphasizes that causality is inferred only through statistical methods rather than directly proven in the lab.

When the authors describe where the strongest signal sits, they point to the IFN (interferon) pathway. However, they also state that the study does not directly validate the mechanism in this work.

The authors close by identifying what comes next: understanding why CXCL9-associated proteins increase alongside age and, crucially, whether interrupting that axis can slow the aging-linked immune dysfunction.

One-sentence conclusion for this section: CXCL9-linked aging signatures show up similarly across sexes, and “epigenetically older” cells respond poorly to pathogens with weakened IFN-γ and IL-22 output.

The study’s practical implication is that the CXCL9–TNF inflammatory axis may define a terrain where epigenetic aging clocks accelerate, turning biomarker causality into a more actionable target for intervention development—especially in the EAA_GrimAge window where the causal signal is clearest.