The new systematic review and network meta-analysis, published in Frontiers in Physiology by researchers from Capital University of Physical Education and Sports, Anhui Polytechnic University, and Changzhou University, synthesized 82 randomized controlled trials involving 6,493 adults with a mean age of 71.04 years. The team compared four structured exercise modalities: aerobic training (AT), resistance training (RT), combined aerobic-plus-resistance training (AT+RT), and multicomponent exercise (MCEx), which integrates three or more training types including balance or flexibility work. Their outcomes of interest were three core inflammatory biomarkers: interleukin-6 (IL-6), C-reactive protein (CRP), and tumor necrosis factor-alpha (TNF-α), the molecular fingerprints of inflammaging.

Methodologically, the study went beyond a conventional meta-analysis. The authors first ran a three-level multilevel meta-analysis to handle a problem most prior syntheses ignored: the statistical dependency created by multi-arm trials, in which one study compares several exercise groups against a shared control. Treating those correlated effect sizes as independent artificially inflates precision and biases estimates. They then performed a frequentist random-effects network meta-analysis, combining direct and indirect evidence to rank all four modalities simultaneously, using the SUCRA statistic, the surface under the cumulative ranking curve, to express each treatment’s probability of being the most effective. Finally, a double subgroup analysis stratified results by both exercise type and participants’ health status, distinguishing healthy older adults from those living with chronic diseases such as hypertension, type 2 diabetes, sarcopenia, and COPD.

The headline result is encouraging. Compared with control conditions, structured exercise significantly reduced all three biomarkers: IL-6 fell with a standardized mean difference (SMD) of -0.31, CRP fell with an SMD of -0.44, and TNF-α fell with an SMD of -0.40, all statistically significant. In plain terms, exercise training acts as a genuine, measurable anti-inflammatory intervention in people over 60, producing effects of small-to-moderate magnitude across a heterogeneous pool of trials lasting from 4 to 72 weeks, most commonly 12 weeks at three sessions per week.

The comparative rankings are where the study gets interesting. In the network meta-analysis, aerobic training ranked first for both IL-6 (SUCRA 84.6%) and TNF-α (SUCRA 97.5%, with a striking 90.6% probability of being the single best intervention). Aerobic training also produced statistically significant reductions versus control on all three markers, with the TNF-α effect (SMD = -0.97) the largest observed. Resistance training, by contrast, ranked first for CRP (SUCRA 89.1%, 65.1% probability of being best), with a significant effect versus control (SMD = -0.68), but no significant effects on IL-6 or TNF-α. Combined training and multicomponent exercise showed positive trends but did not reach statistical significance against control in the network model, and MCEx consistently sat at or near the bottom of the rankings.

The double subgroup analysis added a clinically important layer of nuance. For older adults with chronic diseases, resistance training and combined training significantly improved IL-6 (SMDs of -0.685 and -0.531, respectively) and CRP levels, suggesting that loading the muscles may be particularly valuable when an elevated inflammatory burden is already present. For healthy older adults, the opposite pattern emerged for TNF-α: aerobic training was the only modality with a statistically significant effect (SMD = -0.872), while none of the four modalities reliably lowered TNF-α in the chronic disease group. Health status, in other words, may modify which exercise stimulus delivers the strongest anti-inflammatory return, a finding no previous meta-analysis had systematically tested.

Why would different modalities target different molecules? The authors point to plausible mechanisms rather than proven pathways, since the included trials did not measure molecular signaling directly. Aerobic training is thought to dampen the nuclear factor kappa B (NF-κB) pathway, the master transcriptional switch for pro-inflammatory genes, in part by up-regulating microRNA-146a, a negative regulator that suppresses key adapter molecules in innate immune signaling. Resistance training may work through a different route: reshaping body composition, shrinking the adipose tissue depots that secrete inflammatory cytokines, and boosting anti-inflammatory cytokines such as IL-10, thereby improving the IL-10-to-TNF-α ratio. CRP, produced by the liver largely in response to IL-6, may respond especially well to interventions that alter metabolic and adiposity-related signals, which could explain resistance training’s edge on that marker.

Before anyone rewrites their exercise prescription, however, the authors are emphatic about the caveats. Under the GRADE framework, the certainty of evidence for all three outcomes was rated very low. Of the 82 trials, 53 carried an overall high risk of bias, mostly because blinding participants and trainers in exercise studies is notoriously difficult, and only 14 trials were judged low risk. Statistical heterogeneity was high across the board (I² of 82% for IL-6, 75.5% for CRP, and 88.8% for TNF-α), and funnel-plot tests suggested publication bias for IL-6 and CRP. Notably, trim-and-fill analysis imputed missing studies and produced even larger effect estimates, meaning the bias does not reverse the overall direction of the findings. But the ranking numbers deserve humility: aerobic training’s 54.7% probability of being best for IL-6 leaves roughly a 45% chance another modality is truly superior.

The study is candid about its own discrepancies too. The multilevel subgroup analysis suggested combined training was the strongest modality for IL-6, while the network meta-analysis put aerobic training on top. The authors attribute this to the different analytical frameworks: the network model exploits indirect comparison paths through shared controls and may better integrate all available information, whereas traditional subgroup analysis uses only direct comparisons. Reassuringly, formal inconsistency tests and node-splitting found no significant conflict between direct and indirect evidence, and sensitivity analyses varying the assumed within-study correlation and removing one study at a time left the conclusions essentially intact.

The bottom line is exploratory but actionable in spirit. All active exercise modalities beat no exercise at lowering inflammaging markers, which is the most robust conclusion in the dataset. The rankings hint that aerobic training is the most probable best choice for IL-6 and TNF-α, resistance training for CRP, and that older adults with chronic disease may gain the most from resistance-based or combined programs, while multicomponent exercise shows no consistent anti-inflammatory advantage despite its other well-documented benefits for balance and function. With the authors stressing that no definitive clinical recommendations can yet be made, they call for larger, rigorously blinded trials with standardized protocols, head-to-head modality comparisons, and dose-response analyses. Until then, the message to an aging population is familiar but now better quantified: keep moving, and let the biomarkers decide the details.

Subject of Research: Comparative anti-inflammatory effects of exercise training modalities on IL-6, CRP, and TNF-α in older adults

Article Title: Comparative efficacy of aerobic training, resistance training, combined training, and multicomponent exercise on interleukin-6, C-reactive protein, and tumor necrosis factor-α levels in older adults (≥60 years): a systematic review and network meta-analysis

Article References: Li, W., Chu, R., Xie, Y., Du, Y., & Ni, T. (2026). Comparative efficacy of aerobic training, resistance training, combined training, and multicomponent exercise on interleukin-6, C-reactive protein, and tumor necrosis factor-α levels in older adults (≥60 years): a systematic review and network meta-analysis. Frontiers in Physiology, 17, Article 1837579. https://doi.org/10.3389/fphys.2026.1837579

Image Credits: AI Generated

DOI: 10.3389/fphys.2026.1837579

Keywords: inflammaging, aerobic training, resistance training, network meta-analysis, interleukin-6, C-reactive protein, TNF-alpha, older adults, systematic review, exercise physiology, chronic inflammation, NF-kappaB