The research team, led by Zhengxin Hui and Liang Li, who contributed equally to the work, and supervised by Yinping Zhang and Tonggang Fan, conducted systematic searches of PubMed, Web of Science, Embase and the Cochrane Library up to August 2026. From these searches they assembled an unusually large evidence base: 149 randomized controlled trials encompassing 13,004 patients with cancer. Randomized controlled trials are considered the gold standard for estimating causal effects because participants are assigned by chance to an intervention or a comparison group, which balances out the confounding factors that plague observational studies. Pooling 149 of them therefore offers one of the most comprehensive pictures to date of how different exercise prescriptions perform in cancer populations.

What makes this study methodologically distinctive is its dual approach. First, the authors performed a Bayesian network meta-analysis, a technique that goes beyond traditional pairwise meta-analysis by allowing every intervention to be compared with every other intervention, both directly, when trials tested two modalities against each other, and indirectly, when trials linked them through a common comparator such as usual care. This creates a network of evidence in which the relative standing of each exercise type can be estimated even if no single trial ever compared them all in one experiment. To rank the interventions, the team used the surface under the cumulative ranking curve, or SUCRA, a metric that runs from zero to 100 percent and expresses the probability that a given treatment ranks highest among all the options evaluated.

The SUCRA rankings produced a clear shortlist of top performers. Qigong and Tai Chi, the slow, flowing mind-body practices rooted in traditional Chinese movement culture, emerged among the highest-ranked interventions, alongside high-intensity interval training, better known as HIIT, and combined aerobic and resistance training. The prominence of Qigong and Tai Chi is striking because these are low-impact, low-load practices that many patients with advanced disease, fatigue or treatment-related limitations can actually perform. Their high ranking suggests that the gentle, rhythmic, breath-coordinated character of these practices may deliver quality-of-life benefits that rival far more strenuous regimens, a finding with immediate relevance for patients who cannot tolerate intense exercise.

HIIT, by contrast, represents the opposite end of the intensity spectrum. It alternates short bursts of near-maximal effort with recovery periods, and it has attracted growing interest in exercise oncology because it achieves substantial cardiovascular and muscular stimulus in a compressed amount of time. Combined aerobic and resistance training, the third top-ranked modality, blends endurance work such as walking or cycling with strength work using weights or bodyweight resistance. The convergence of these three very different modalities at the top of the rankings underscores an important point: there is no single best exercise for quality of life in cancer care, but rather several distinct pathways, each with its own dose requirements, that appear to lead to similar destinations.

The second analytical pillar of the study was a network dose-response meta-analysis, abbreviated MBNMA in the paper, which models how the effect of an intervention changes as a function of its dose. The researchers quantified exercise dose using MET-minutes per week, a standard unit in exercise science in which one metabolic equivalent, or MET, represents the energy cost of resting quietly, and brisk physical activity typically registers at three to eight METs. A person walking briskly for five hours a week at roughly four METs would accumulate about 1,200 MET-minutes. Rather than assuming that more exercise is always better, the team fitted an Emax model, a nonlinear function borrowed from pharmacology that describes how a response rises with dose, approaches a maximum, and then plateaus.

The dose-response modelling delivered one of the study’s headline findings: the association between exercise dose and quality of life is nonlinear, with the model-estimated overall effect peaking at approximately 2,000 MET-minutes per week. Below that threshold, additional exercise appears to buy meaningful gains in wellbeing; beyond it, the model suggests diminishing returns. This kind of plateau is biologically plausible. Quality of life is constrained not only by physical capacity but by fatigue, psychological distress, sleep and social functioning, and once exercise has extracted the improvements it can deliver in these domains, pushing the dose higher may add strain without adding benefit. The finding echoes the shape of dose-response curves reported in other exercise-oncology research and gives clinicians a concrete numerical anchor for prescription.

Crucially, the peak dose was not the same for every modality. The model-estimated maximum effect for Qigong and Tai Chi occurred at a much lower dose, around 750 MET-minutes per week, which corresponds to only a few hours of gentle practice spread across the week. HIIT reached its model-estimated maximum at about 800 MET-minutes per week, consistent with the idea that high-intensity work delivers its stimulus efficiently. Combined aerobic and resistance training required the largest dose, with the model placing its maximum effect at roughly 2,000 MET-minutes per week, in line with the overall peak. In practical terms, a patient who enjoys and tolerates combined training may need to invest considerably more weekly activity to reach the modelled optimum, whereas a patient drawn to mind-body practice or interval work may achieve comparable modelled benefit at a fraction of that volume.

The authors are careful to frame these estimates as model-derived values that require prospective validation. Network meta-analysis and dose-response modelling are observational in the sense that they reanalyse existing trial data rather than testing prescriptions in new patients, and the Emax parameters depend on the doses represented in the included trials, which were not designed with a common dose grid in mind. Heterogeneity across cancer types, treatment stages, outcome instruments and intervention durations can all influence the pooled estimates, and the Bayesian framework makes its assumptions explicit but does not eliminate them. The study was funded by the Shanghai Oriental Talent Program Youth Project, and the authors declare no competing interests. The work is published open access, making the full analysis available to clinicians and researchers worldwide.

Even with those caveats, the study arrives at a moment when exercise is being formally integrated into cancer care pathways in many health systems, and it offers something those pathways have lacked: a comparative, quantified map of the options. For a patient beginning rehabilitation after chemotherapy, the findings suggest that a low-dose Qigong or Tai Chi routine may be an evidence-supported starting point, particularly when fatigue or deconditioning makes vigorous activity unappealing. For a fitter patient, HIIT or a combined aerobic and resistance program, dosed with the modelled peaks in mind, represents an alternative route to the same goal. What the analysis ultimately reinforces is that in exercise oncology, as in pharmacology, the prescription matters, the dose matters, and the best regimen is the one that matches the right modality to the right patient at the right intensity.

Subject of Research: Comparative effectiveness and dose-response of exercise interventions on quality of life in patients with cancer

Article Title: Optimizing exercise regimens for quality of life in patients with cancer: a network meta-analysis and dose–response study

Article References: Hui, Z., Li, L., Zhang, Y., & Fan, T. (2026). Optimizing exercise regimens for quality of life in patients with cancer: a network meta-analysis and dose–response study. BMC Cancer. https://doi.org/10.1186/s12885-026-17109-5

Image Credits: AI Generated

DOI: 10.1186/s12885-026-17109-5

Keywords: cancer rehabilitation, exercise oncology, quality of life, network meta-analysis, dose-response, HIIT, Tai Chi, Qigong, aerobic training, resistance training, randomized controlled trials, SUCRA

The research team, led by Ted G. Graber of East Carolina University together with colleagues including Drew Theobald, Srinivas Sriramula, and Lisandra de Castro Braz, set out to fill a conspicuous gap in the exercise-aging literature. Most preclinical studies of exercise have focused on male animals, even though males and females age along measurably different trajectories. The team’s own prior longitudinal work showed that female mice lose four-limb strength faster than males in midlife, gain excessive body fat as they age, and follow distinct patterns of functional decline. Whether a demanding interval program could protect aging females across physical, cognitive, and cardiovascular domains simultaneously had never been tested in such depth.

The experiment was elegantly simple in design. Twenty-one virgin female C57BL/6J mice were aged to 20 months—roughly the mouse equivalent of late middle age into older adulthood—before the intervention began. After baseline testing, eighteen mice were randomized into two groups of nine: one group underwent fourteen weeks of progressively harder HIIT on a treadmill, while the sedentary controls experienced a sham treatment, spending identical time on the stationary treadmill with only brief, slow belt movement. This sham was critical, because it matched the environmental stimulation and handling of the trained animals without providing any actual exercise stimulus, allowing the researchers to attribute differences to training itself rather than to novelty or human contact.

The training protocol was individually personalized and progressive. Each mouse’s interval speeds were anchored to its own maximum treadmill speed, measured before training and again at the seven-week midpoint. Sessions began with a two-minute warm-up walk, then alternated thirty-second ramps up to a target percentage of maximum speed, sixty seconds of sustained running, and active recovery at walking pace. Over fourteen weeks the animals advanced from three intervals per session to seven, with peak intervals reaching 80 percent of each mouse’s maximum speed. By the end of the study, the trained mice had significantly increased their maximum running speed—a clear sign of genuine aerobic adaptation—while the sedentary group showed no such improvement.

The physical function results were dramatic. Aerobic capacity, measured by a maximum-speed treadmill test, surged by 71 percent in the trained mice while remaining flat in controls. Four-limb strength and endurance, assessed with an inverted cling test in which mice hang upside down from a wire grid, fell by 67 percent in sedentary animals but only 28 percent in the trained group. Forelimb grip strength declined 16 percent in sedentary mice yet was fully maintained with training. Overall motor function on the rotarod, a rotating rod that tests balance, coordination, and power, improved by 39 percent in the exercised animals. To capture the big picture, the team summed these measures into their validated composite score, the Comprehensive Functional Assessment Battery, or CFAB. Sedentary mice declined significantly on this composite; trained mice held steady.

Cognition told an equally compelling story. The researchers introduced a new composite instrument, the Cognitive Assessment Battery, combining open-field exploration, novel object recognition, the Y-maze spontaneous alternation test, and the puzzle box, a task that probes executive function and problem-solving by requiring mice to navigate increasingly difficult barriers to reach a dark escape box. Trained mice cut their total puzzle-box completion time by 73 percent, compared with 40 percent in sedentary animals, and improved dramatically on the hardest blocked-exit challenge. In the open field, they traveled 89 percent farther and spent twice as much time in the center of the arena, a sign of reduced anxiety-like behavior. Y-maze spontaneous alternations, a measure of working and spatial memory, rose 61 percent in the trained group. The composite CAB score increased 34 percent with training, a significant gain over controls. Only long-term recognition memory, measured by novel object discrimination after 24 hours, showed no difference—a result the authors attribute to the test’s limited sensitivity in non-pathological aging rather than to a true absence of benefit.

Perhaps the most clinically resonant finding concerned frailty, the syndrome of accumulated physiological deficits that erodes independence in older adults. At baseline, none of the mice were frail. By the end of the study, four of nine sedentary mice had progressed to frank frailty and two more were pre-frail, leaving only two animals classified as non-frail. In the trained group, not a single mouse became frail: four remained non-frail and four were pre-frail. Strikingly, the team confirmed that CFAB scores track frailty status closely—all pre-frail mice scored below a CFAB threshold of minus two, and all frail mice below minus six—validating the composite physical function score as a practical proxy for frailty in females, extending earlier work in males.

The cardiovascular data added further weight. Systolic blood pressure, measured noninvasively with tail-cuff plethysmography, fell by 17 mmHg, or 12.5 percent, in trained mice, while diastolic pressure dropped 15 percent and mean arterial pressure fell 14 percent. Although the small sample size meant between-group comparisons reached only large effect sizes rather than statistical significance, the within-group reductions were robust. At the tissue level, the differences were even more vivid. Sedentary aged hearts accumulated significantly more collagen—the fibrotic scar tissue that stiffens the ventricle and predisposes older hearts to failure—while the trained hearts showed collagen levels indistinguishable from those of young adult mice. Fibroblast activation, marked by alpha-smooth muscle actin expression, rose significantly in sedentary hearts but only modestly in trained ones, and echocardiography hinted that training blunted the age-associated thickening of the ventricular wall.

Inside the brain, the story centered on neuroinflammation. In the hippocampal CA1 region, a hub for learning and memory that is especially vulnerable to aging, sedentary mice showed elevated expression of the inflammatory cytokine IL-1β, along with increased IBA1 marking microglial activation and elevated GFAP in astrocytes. HIIT significantly reduced IL-1β expression relative to sedentary controls and nudged glial cells toward a less reactive morphology, with trends toward increased microglial branching and reduced astrocytic hypertrophy. Because IL-1β is a central driver of age-related synaptic dysfunction and impaired plasticity, its suppression may represent one of the earliest and most exercise-responsive mechanisms linking interval training to preserved cognition. In skeletal muscle, the soleus shifted toward a slower, more oxidative fiber profile, with type I fibers rising from 57 to 64 percent, while the remaining type IIa fibers grew substantially larger—a 47 percent increase in cross-sectional area with a strong effect size—suggesting a compensatory strategy that preserves force without sacrificing endurance.

The authors are careful about the limits of the work: a single strain, a single age group, a modest sample size, and one sex at one endpoint. Yet the breadth of protection observed in a fourteen-week program begun late in life is remarkable, and the study is, to the team’s knowledge, the first comprehensive evaluation of HIIT in older female mice. The broader implication is provocative. If intense interval exercise can simultaneously defend muscle, brain, and heart, and hold frailty at bay, it behaves less like a fitness routine and more like a genuine geroscience intervention—one that targets multiple hallmarks of aging, including inflammaging, tissue fibrosis, and metabolic dysfunction. The next frontier is mechanistic: identifying the circulating exerkines and extracellular vesicles that carry exercise’s signals between organs, knowledge that could one day yield exercise-mimetic drugs for people who cannot train. For now, the message from these aged mice is clear and energizing: it is never too late to start, and the harder intervals may pay the biggest dividends.

Subject of Research: Effects of high-intensity interval training on physical, cognitive, and cardiovascular aging in female mice

Article Title: High-intensity interval training in aged female mice preserves physical, cognitive, and cardiovascular function

Article References: Theobald, D., Williamson, P., Johnston, A., Tripp, L., Olabiyi, A. A., Silvers, X., Dickerson, A., Tran, T. D., de Castro Braz, L., Sriramula, S., & Graber, T. G. (2026). High-intensity interval training in aged female mice preserves physical, cognitive, and cardiovascular function. GeroScience. https://doi.org/10.1007/s11357-026-02489-0

Image Credits: AI Generated

DOI: 10.1007/s11357-026-02489-0

Keywords: HIIT, aging, frailty, exercise, female mice, cognition, cardiovascular health, cardiac fibrosis, neuroinflammation, sarcopenia, GeroScience, healthspan

The trial enrolled 41 middle-aged and older adults, randomly assigning them to receive either a daily supplement delivering roughly 700 milligrams of polyphenols or an identical placebo. After a 30-day loading phase without exercise, participants began a carefully supervised 12-week programme: two full-body resistance training sessions per week, plus a third session combining lower-body resistance work with 7 to 9 minutes of high-intensity intervals on a cycle ergometer. Loads progressed from moderate to high intensity throughout, with lower-body exercises moving from approximately 12 to 6 repetition maximum and upper-body work from 15 to 10. Before and after the intervention, researchers took biopsies from the vastus lateralis, the large quadriceps muscle on the side of the thigh, and sequenced the RNA within.

The sequencing pipeline was demanding. Of 79 muscle samples collected, 76 passed quality control, yielding a final dataset of 38 participants — 17 in the polyphenol group and 21 on placebo. The team aligned trimmed reads to the human reference genome GRCh38 and applied a stringent statistical framework, adjusting every model for age, sex and, crucially, the estimated proportion of slow- and fast-twitch muscle fibres in each sample. That last adjustment proved essential. Fibre-type composition influenced the expression of 633 genes in the dataset, far more than age, which affected only nine. Sex was an even stronger confounder, shaping 1,040 genes. Without accounting for these factors, apparent training effects could easily have reflected baseline biological differences between individuals rather than true responses to exercise.

When the researchers pooled all participants and compared pre- and post-training samples, they found 60 transcripts that reached statistical significance after correction for multiple testing. But only seven of those also exceeded a fold-change threshold of greater than twofold — the combined statistical and effect-size criterion the team used to flag changes of potentially larger biological magnitude. Among those seven, six were upregulated and one was downregulated. The largest increases appeared in CLLU1 and PRND, while the most pronounced decrease occurred in MYH1, the gene encoding the myosin heavy chain IIx isoform that powers the fastest, most glycolytic muscle fibres.

That MYH1 signal is biologically intriguing. Exercise programmes blending resistance and endurance components are known to nudge muscle toward a more oxidative phenotype, often involving transitions from type IIx toward type IIa fast fibres. Yet the histological analyses previously performed in this same cohort showed that training enlarged type II fibres overall without changing the relative distribution of type I and type II fibres, and did not separately quantify the IIa and IIx subtypes. The authors therefore interpret the MYH1 downregulation as a transcript-level change related to the fast contractile phenotype, not as direct evidence of a fibre-type switch. Similarly, the upregulation of LOX, which encodes lysyl oxidase — an enzyme that cross-links collagen and elastin in the extracellular matrix — fits with the well-documented role of exercise in remodelling the connective tissue scaffolding that transmits force within muscle. But without coordinated enrichment of broader matrix-related pathways, the researchers caution that LOX stands as an individual transcript-level signal rather than proof of systematic extracellular matrix remodelling.

Perhaps the most striking aspect of the findings is what was absent. Despite the robust physiological adaptations documented in this cohort in earlier analyses — gains in muscle mass, strength, aerobic capacity and mitochondrial respiratory capacity — the transcriptional response was predominantly modest in magnitude and failed to converge on any coherent pathway-level signature. No robust enrichment emerged from gene ontology, Reactome or KEGG pathway analyses. This is not as paradoxical as it may seem. Accumulating multi-omic evidence, including large-scale consortium studies, indicates that chronic exercise adaptations in human muscle are typically distributed across many genes with small effect sizes — a coordinated fine-tuning of biological systems rather than a dramatic transcriptional overhaul. Recent single-nucleus RNA sequencing studies have reported similarly limited numbers of differentially expressed genes even with cell-type-specific resolution.

The polyphenol story ended in near silence. At baseline, before any training began, the supplement and placebo groups were nearly indistinguishable at the transcriptomic level, with only three FDR-significant transcripts — and just one, the oxidative stress response gene HMOX1, exceeding the fold-change threshold, and even that with a modest effect. The group-by-time interaction analysis, designed to detect whether the supplement changed how muscle responded to training, identified 14 FDR-significant transcripts, of which eight met the combined criterion. But inspection of individual expression profiles revealed low or sparse expression and marked inter-individual variability rather than consistent within-subject changes. The most biologically interpretable candidate, KERA, which encodes the extracellular matrix proteoglycan keratocan, was expressed at very low levels and appeared without any accompanying response in other matrix-related genes. Heatmaps of top-ranked genes showed no clear clustering by group or timepoint, and no significantly enriched pathways emerged in KEGG analysis.

These molecular null results align seamlessly with the broader picture from the same trial. Polyphenol supplementation had previously failed to augment training-induced improvements in muscle mass, strength and aerobic capacity, did not enhance mitochondrial bioenergetics or redox balance beyond what training alone achieved, did not alter the inflammatory response to exercise, and left skeletal muscle accumulations of advanced glycation end-products unchanged. Taken together, the authors conclude that polyphenol supplementation did not meaningfully modulate the skeletal muscle transcriptional response to concurrent training. For a supplement category often marketed on the premise of priming redox-sensitive and inflammatory signalling pathways, the message is pointed: any modulatory effects, if they exist, are too subtle to register at the level of muscle gene expression or physiological function.

The study also carries methodological lessons for the field. By incorporating in silico estimates of fibre-type composition — derived from a validated gene-signature matrix using the FibeRtypeR tool — as a covariate, the researchers reduced the risk of mistaking baseline phenotypic heterogeneity for intervention effects, a confound that many human muscle transcriptomics studies have ignored. The authors acknowledge limitations, however: the sample size limits statistical power for detecting small but real changes; bulk RNA sequencing cannot resolve cell-type-specific adaptations; transcriptomics misses post-transcriptional and translational regulation, including ribosome biogenesis, a central driver of muscle hypertrophy whose signals are largely invisible to mRNA sequencing; and relatively low RNA quality and a modest 10-nanogram input may have reduced detection sensitivity. The stringent fibre-type adjustment itself likely suppressed the number of detectable differentially expressed genes.

For ageing adults, the practical takeaway remains firmly anchored in the sweat, not the supplement. Twelve weeks of combined resistance training and high-intensity intervals produced measurable, if molecularly subtle, remodelling in the transcriptome of ageing muscle — touching genes of the extracellular matrix and the contractile machinery — while delivering well-documented gains in strength, fitness and mitochondrial function. The polyphenol extract, whatever its merits elsewhere, added no detectable molecular edge. As the population ages and sarcopenia looms as a major driver of frailty and loss of independence, the trial reinforces a simple hierarchy: progressive mechanical loading of muscle remains the most reliable lever we have, and no berry extract has yet been shown to turn its dials further.

Subject of Research: Skeletal muscle transcriptomic responses to concurrent resistance and high-intensity interval training with and without polyphenol supplementation in ageing adults

Article Title: Skeletal Muscle Transcriptomic Responses to Concurrent Training With Polyphenols in Ageing Adults: A Randomized Trial

Article References: Guadalupe‐Grau, A., Gudra, D., Lismane, D., Raudeniece, J., Akuratere, G., Viksne, K., Gailite, L., Vilne, B., Pivovarova‐Ramich, O., Flensted‐Jensen, M., Dela, F., & Reihmane, D. (2026). Skeletal Muscle Transcriptomic Responses to Concurrent Training With Polyphenols in Ageing Adults: A Randomized Trial. Journal of Cachexia, Sarcopenia and Muscle, 17(5), Article e70391. https://doi.org/10.1002/jcsm.70391

Image Credits: AI Generated

DOI: 10.1002/jcsm.70391

Keywords: skeletal muscle, ageing, transcriptomics, RNA sequencing, resistance training, HIIT, polyphenols, randomized trial, extracellular matrix, MYH1, mitochondria, sarcopenia