Exercise training prior to and during cancer in mice preserves muscle mass, reduces tumour weight and suppresses molecular mediators of cachexia

Abstract

Abstract

We investigated the prophylactic effects of exercise before and during cancer cachexia (CC) using a model designed to mimic endurance and resistance (i.e., concurrent) adaptations. Male and female Balb/c mice were randomly assigned to exercise or control groups whereby exercise groups were subjected to an 8‐week voluntary progressive weighted wheel running (PoWeR) programme of habitual loading‐mediated physical activity beginning at 8 weeks of age. At 16 weeks of age, mice were injected bilaterally with colon‐26 adenocarcinoma (C26) cells or phosphate‐buffered saline, and exercise training was maintained throughout disease progression. Twenty‐five days post‐tumour induction, we assessed whole‐body and muscle phenotype, muscle protein synthesis, a priori targeted gene expression, and transcriptomic adaptations via RNA sequencing. PoWeR training preserved skeletal muscle mass across nearly all muscle groups and maintained tumour‐free body and cardiac mass. Muscle mass adaptations related to running volume, and running distance relative to controls were not appreciably reduced by tumour status. Tumour burden was reduced after ∼11.5 weeks of PoWeR compared to sedentary, but this was not explanatory for muscle adaptations. PoWeR induced a faster‐to‐slower muscle fibre type transition in the gastrocnemius and suppressed key protein turnover markers (Redd1, Murf1, Atrogin, Ubc, Gadd45a) as well as the mitophagy‐related marker Bnip3 in tumour‐bearing muscle; 24 h muscle protein synthesis remained stable. PoWeR counteracted tumour‐induced impairments in the muscle mitochondrial‐ and metabolic‐related transcriptome. Collectively, physical activity prior to and during cancer preserves muscle mass, reduces tumour growth and mitigates molecular drivers of CC, underscoring its preventive and therapeutic potential as a lifestyle intervention.

Key points

• Cancer cachexia (CC) is a severe, multifactorial syndrome with limited effective therapies.

• Exercise training has emerged as a promising non‐pharmacological approach to mitigate CC.

• Concurrent endurance and resistance training, initiated prior to and maintained during cancer, preserves skeletal muscle mass and reduces tumour burden in C26 colorectal tumour‐bearing mice.

• Concurrent exercise training suppresses key mitochondrial‐ and metabolic‐related molecular mediators of CC.

• Concurrent exercise training may serve as a preventive and therapeutic non‐pharmacological strategy against CC.

Abstract figure legend PoWeR training (11.5 weeks of voluntary wheel running with progressively increased resistance) before and during cancer attenuates muscle loss, limits tumour growth, drives a transition toward a more oxidative muscle phenotype (IIB‐to‐IIA shift), downregulates cachexia‐associated pathways and enhances mitochondrial performance in C26 tumour‐bearing mice. Created with BioRender.com.

Introduction

Cancer cachexia (CC) is a multifactorial syndrome defined by unintentional loss of body weight and muscle mass in patients (Argiles et al., 2014). Cachexia affects up to 80% of all cancer patients and accounts for as much as 40% of cancer‐related deaths (Argiles et al., 2014; Fearon et al., 2011). Despite numerous proposed strategies primarily involving nutritional or pharmacological approaches, current interventions have been of limited success, leaving a critical gap in the treatment and recovery of patients with CC. Exercise training has emerged as a potential therapeutic approach for CC (Courneya et al., 2000; Tsitkanou et al., 2022), though its application must be carefully tailored to avoid contraindications related to comorbidities (Argiles et al., 2012). Little is known about the role of lifestyle or long‐term physical activity earlier in life (i.e., exercise pre‐conditioning) in preventing or delaying CC, even though such strategies could be crucial for improving survival and quality of life in cancer patients.

The heterogeneity of preclinical exercise studies in CC makes it challenging to draw definitive conclusions about the utility of exercise in disease development and progression. Nevertheless, combined resistance and endurance exercise (concurrent exercise) at moderate intensities appears particularly effective in counteracting CC (Ranjbar et al., 2019; Tsitkanou et al., 2022; Wolin et al., 2012), a recommendation endorsed by the American College of Sports Medicine for cancer patients (Wolin et al., 2012). A meta‐analysis showed that exercise (aerobic, resistance and concurrent exercise grouped in the analysis) initiated after tumour inoculation does not prevent cancer‐induced muscle atrophy in rodents (Niels et al., 2020). In line with this, our previous work shows short‐term progressive weighted wheel running (PoWeR) for 3.5 weeks beginning concurrently with cancer in C26 colorectal tumour‐bearing mice did not attenuate cancer‐associated muscle loss (Tsitkanou et al., 2026). However, the adaptive response to exercise training was largely preserved during CC in both sexes, including fibre‐type remodelling in limb muscles and cardiac hypertrophy in both sexes (Tsitkanou et al., 2026). Consistent with our findings, Collao et al. (2023) reported that initiating PoWeR following rhabdomyosarcoma induction and its treatment with chemotherapy and radiation preserved muscle mass while partially reversing cancer‐associated inflammatory and fibrotic transcriptional changes.

Contrasting with exercise during cancer, long‐term exercise (encompassing all types of exercise) initiated prior to symptom onset, mimicking a physically active lifestyle, reduces the risk of CC and can delay its onset (Niels et al., 2020). Ranjbar et al. (2019) reported that 5.5 weeks of concurrent training (4 weeks prior to and 1.5 weeks following tumour implantation), consisting of resistance exercise (inclined ladder climbing with progressive load) combined with aerobic training (25 min of wheel running at 5–9 m/min) prevented tumour‐induced muscle wasting and weakness, accompanied by attenuated autophagy induction and restoration of muscle succinate dehydrogenase (SDH) activity. Collectively, these findings point to the timing of exercise being a key factor determining its effectiveness in the context of CC and highlight the potential of concurrent exercise to counteract muscle wasting and mitigate molecular dysregulation during tumour progression. Still, the potential benefits of a high level of physical activity prior to and during tumour progression remain understudied.

In this study, we aimed to investigate the effects of long‐term exercise pre‐conditioning, which was maintained post‐tumour implantation, designed to reflect lifestyle physical activity, as a preventive strategy against CC. Our results demonstrate that long‐term PoWeR, performed prior to C26 cell implantation and maintained during tumour progression, led to a reduction in tumour weight, preservation of skeletal muscle mass across nearly all muscles, and maintenance of tumour‐free body mass. Moreover, this exercise intervention attenuated markers of protein degradation, contributed to the restoration of the muscle mitochondrial transcriptome, and promoted a faster‐to‐slower muscle fibre‐type transition. Overall, these findings indicate that long‐term exercise pre‐conditioning continued through cancer progression may represent a viable approach to the prevention of cancer‐induced muscle wasting.

Methods

Ethical approval

All animal procedures were approved by the Institutional Animal Care and Use Committee of the University of Arkansas (AUP 22035) and were conducted in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals (National Research Council, 2011). The authors confirm that all experiments comply with the ethical principles under which The Journal of Physiology operates and adhere to the ARRIVE guidelines for reporting in vivo experiments.

Animals and experimental design

Sixty‐two Balb/c mice (n = 32 males, n = 30 females) were obtained from The Jackson Laboratory (Bar Harbor, ME, USA; stock no.: 001026) and housed in a temperature‐controlled environment (∼22°C) under a 12:12‐h light–dark cycle. Mice had ad libitum access to standard rodent chow and water throughout the study.

At 6 weeks of age, mice arrived at the animal facility and were acclimatized for 1 week. Thereafter, half of the animals were randomly assigned to exercise groups and single‐housed in wheel‐running cages with unloaded wheels (no resistance) for a 7‐day familiarization. At 8 weeks of age, mice began an 8‐week voluntary PoWeR training programme. Resistance was gradually increased from 2 g to 6 g as follows: 2 g (week 1), 3 g (week 2), 4 g (week 3), 5 g (weeks 4–5) and 6 g (weeks 6–8). At 16 weeks of age, all mice were bilaterally injected subcutaneously with either Colon‐26 Carcinoma (C26, National Cancer Institute, Bethesda, MD, USA) cells (total cell volume of 1 × 10⁶) suspended in 200 µL sterile phosphate‐buffered saline (PBS) (i.e., 5 × 10⁵ cells suspended in 100 µl sterile PBS for each flank) or equal volume of sterile PBS as a sham control. Tumours were allowed to develop for 25 days, which defined the experimental endpoint based on recent studies using this model (Cabrera et al., 2023; Delfinis et al., 2022; Tsitkanou et al., 2024, 2026). Following C26/PBS injection, mice in exercise groups continued voluntary PoWeR training with a constant 6 g resistance for an additional 24 days (Fig. 1). Running distances (km/day) were recorded using ClockLab software (Actimetrics, Wilmette, IL, USA). Sedentary groups remained housed without wheel access throughout the study.

Figure 1. Experimental design of the study.

Created with BioRender.com.

Our previous findings using the PoWeR model (Tsitkanou et al., 2026) and current results demonstrated few sex‐specific differences; therefore, data from males and females were combined for all subsequent analyses. Four experimental groups were included: (i) sedentary mice injected with PBS (PBS SED; n = 22 total: 12 males, 10 females), (ii) PBS‐injected mice performing PoWeR training (PBS PoWeR; n = 11 total: 6 males, 5 females), (iii) sedentary mice injected with C26 cells (C26 SED; n = 18 total: 8 males, 10 females), and (iv) C26‐injected mice performing PoWeR training (C26 PoWeR; n = 11 total: 6 males, 5 females).

Body weight (g), food intake (g/day) and running distance (km/day) were recorded weekly. Area under the curve (AUC) was calculated from weekly measurements (weeks 1–12) for running distance and food consumption. Since running distance tended to decrease following tumour inoculation, AUC was also calculated separately for the periods before and after C26 cell inoculation. At 24 days post‐C26/PBS injection, mice underwent grip strength testing and subsequently received intraperitoneal injections of 99.9% deuterium oxide (D₂O) for assessment of 24‐h fractional protein synthesis rates (see ‘Deuterium administration and 24‐h fractional protein synthetic rate’). Wheel‐running cages were subsequently locked 24 h prior to tissue collection to allow washout/recovery from the acute effects of exercise (Fig. 1). At the endpoint (25 days post‐C26 inoculation; morning hours: 08.00–12.00 h), skeletal muscles from the hindlimbs, liver, spleen, heart, gonadal fat and tumours were collected, weighed, snap‐frozen and stored at −80°C for subsequent analyses. All tissue weights were normalized by tibia length as a surrogate of body size at endpoint. As not all mice performed wheel running uniformly, inclusion criteria for the final exercise groups excluded animals that failed to run (i.e., <1 km/day for more than two consecutive weeks). These mice were euthanized and excluded from further analysis (1 mouse from PBS PoWeR and 1 mouse from C26 PoWeR), as they could not be classified as either exercise or sedentary controls. In addition, mice with tumour weights <0.3 g were excluded from the final analyses, as they did not exhibit cachexia (4 mice from C26 PoWeR).

Grip strength assessment

Grip strength was assessed in the morning, approximately 24 h before tissue collection (endpoint) and immediately prior to intraperitoneal injection of 99.9% D₂O. Mice were positioned on the grasping grid of a grip strength meter (Harvard Apparatus, Holliston, MA, USA) and gently pulled by the tail until they released the grid, generating a force recorded by the transducer. Each animal underwent three sets of three trials (3 × 3; nine trials total), with a 3–5 min rest period between sets. For each set, the median force value was calculated, and the mean of the three medians was used as the animal's grip strength score. This value was normalized to body weight, and normalized scores were used for the final analysis. A familiarization session using the same protocol (3 × 3 trials) was conducted 1 week prior to testing. All assessments were performed by the same investigator to minimize inter‐experimenter variability.

RNA extraction and targeted mRNA analysis

Quantitative real‐time PCR was performed a priori to assess canonical markers of protein turnover and mitochondrial regulation previously established by us and others to be affected during cachexia to validate mRNA level impacts of cachexia (Cabrera et al., 2023; Delfinis et al., 2022; Tsitkanou et al., 2024). Total RNA was isolated from gastrocnemius muscle using TRIzol (Thermo Fisher Scientific, Waltham, MA, USA, cat. no. 10296028) and the PureLink™ RNA Mini Kit (Thermo Fisher Scientific, cat. no. 12183025) and quantified with a Take3 micro‐volume plate reader (BioTek Instruments, Winooski, VT, USA), as previously described (Cabrera et al., 2023; Lim et al., 2022; Tsitkanou et al., 2024, 2026). cDNA was synthesized using 4 µl of SuperScript IV VILO Master Mix (Thermo Fisher Scientific, cat. no. 11756500) with RNA (∼50 ng/µl) in a total volume of 21 µl. Reverse transcription was performed at 25°C for 10 min, 50°C for 10 min and 85°C for 5 min. cDNA was diluted 1:100 with ddH₂O for storage at −20°C and further diluted to 0.5 ng/µl for RT‐PCR; stock and diluted samples were stored at −80°C and −20°C, respectively (Cabrera et al., 2023; Lim et al., 2022; Tsitkanou et al., 2024, 2026).

Quantitative real‐time PCR (Thermo Fisher Scientific) was performed using 10× TaqMan™ Fast Advanced Master Mix (Thermo Fisher Scientific, cat. no. 4444558), 1× TaqMan™ Gene Expression Assays (FAM, Thermo Fisher Scientific, cat. no. 4331182) and 1× of DNase/RNase‐free water and 8× cDNA sample (concentration of 0.5 ng/µl) to attain a final volume reaction of a 25 µl, as previously described (Cabrera et al., 2023; Tsitkanou et al., 2024, 2026). Cycling conditions were 50°C for 2 min, 95°C for 10 min, followed by 45 cycles of 95°C for 15 s and 60°C for 1 min. 18S (Mm03928990_g1) was used as the housekeeping gene. Target genes included Atrogin1 (Fbxo32, Mm00499523_m1), Murf1 (Trim63, Mm01185221_m1), Ubc (Mm02525934_g1), Gadd45a (Mm00432802_m1), Deptor (Mm01195339_m1), Redd1 (Ddit4, Mm00512504_g1), Opa1 (Mm01349707_g1), Bnip3 (Mm01275600_g1) and Pgc1a (Mm00447183_m1). mRNA levels were quantified using the ΔΔC _t method with 18S as control. Cycle threshold values of 18S were assessed statistically and did not differ between experimental conditions. Relative expression was calculated as $2^{-\mathrm{\Delta }\mathrm{\Delta }{C}_{\mathrm{t}}}$, and fold changes were determined by normalizing each sample to the average of healthy controls as described by our previous studies (Cabrera et al., 2023; Greene et al., 2015; Tsitkanou et al., 2024, 2026). Fold change values were used for statistical analyses.

Deuterium administration and 24‐h fractional protein synthetic rate

Mice were administered an intraperitoneal injection of 99.9% deuterium oxide (D₂O; 20 µl/g body weight; cat. no. 151882‐1L, MilliporeSigma, Burlington, MA, USA) 24 h prior to tissue collection (Tsitkanou et al., 2024, 2026). This dosing strategy achieves ∼2% body water enrichment, sufficient for determining fractional protein synthetic rate (FSR). Following injection, drinking water was supplemented with 4% D₂O to maintain plasma enrichment (Gasier et al., 2009). A 24‐h labelling period was chosen to capture protein synthesis across a complete light–dark cycle while minimizing potential label recycling.

FSR in the gastrocnemius muscle was quantified by gas chromatography–mass spectrometry (GC–MS; 7890A and 5977A, Agilent Technologies, Santa Clara, CA, USA), as previously described by our laboratory (Cabrera et al., 2023; Lim et al., 2022; Tsitkanou et al., 2024, 2026). Peak abundances of ions 99 (alanine) and 100 (deuterated alanine) were extracted from chromatograms, and the 100/99 ion ratio was used to calculate protein‐bound alanine enrichment based on a regression curve generated from [²H]alanine standards (r ² = 0.999). Data were normalized to plasma D₂O enrichment, which was determined from plasma samples collected at harvest and analysed as previously described (Cabrera et al., 2023; Lim et al., 2022; Tsitkanou et al., 2024, 2026).

Immunohistochemistry and image analysis

At tissue collection, gastrocnemius muscles were dissected and mounted in optimal cutting temperature (OCT) medium, then snap‐frozen in liquid nitrogen‐cooled isopentane and stored at −80°C until later analyses. Fibre cross‐sectional area (CSA) and myosin heavy chain (MyHC) fibre‐type analyses on the gastrocnemius were performed as previously described (Murach et al., 2020; Tsitkanou et al., 2026). Sections 8 µm thick were cut at the mid‐belly using an Epredia Cryostar (Portsmouth, NH, USA) NX50 cryostat and allowed to air dry for at least 1 h. Muscle sections were encircled in a hydrophobic Polymerase Activity Probe pen barrier (ImmEdge; Vector Laboratories, Newark, CA, USA) and incubated with primary antibodies for dystrophin (1:100, ab15277; Abcam, Waltham, MA, USA) and MyHCs I, IIA (1:100; BA‐D5, SC‐71; Developmental Studies Hybridoma Bank, Iowa City, IA, USA) for a minimum of 4 h in a PBS cocktail at 4°C. MyHC IIX and IIB were left unstained. After three rounds of PBS washes, slides were incubated in a PBS cocktail of isotype‐specific secondary antibodies conjugated to different fluorescent tags at room temperature for 60–90 min (1:200; AF555, cat. no. 21426; AF488, cat. no. 21121; AF647, cat. no. 21 242; Thermo Fisher Scientific). Following PBS washes, slides were mounted using a 50:50 solution of PBS and glycerol. Images were captured as whole‐muscle cross‐sections at ×20 magnification using a Zeiss AxioImager M2. Fibre CSA and fibre‐type distribution were analysed using MyoVision 2.0, as previously described (Viggars et al., 2022). An average of 6404 ± 1947 fibres (minimum >2000) were analysed. Not all cross‐sections were of suitable quality for immunohistochemical analysis due to freeze–fracture artifacts, so they were excluded. These procedures were performed in an unbiased manner by an investigator not aware of experimental conditions.

Gastrocnemius sections (8 µm) were also used for SDH staining. Sections were air‐dried and incubated in a succinate buffer containing 50 mM sodium phosphate (pH 7.6), 50 mM sodium succinate and 0.5 mg/ml nitroblue tetrazolium for 45 min at 37°C. After incubation, sections were rinsed in deionized water and then processed for fibre typing using the protocol described above. Image analysis was conducted using MyoVision v2.9.1, which simultaneously assesses fibre type and SDH staining intensities in a fibre type specific manner using brightfield and fluorescence microscopy images taken concurrently. Pixel intensities for each fibre were exported and analysed using 8‐bit greyscale (0–255), where black pixels are given a value of zero (arbitrary units) representing darker SDH staining and lighter pixels are given higher values representing weaker staining. For classification of SDH staining intensity, fibres were binned using predefined arbitrary thresholds: 0–100 for dark staining, 101–175 for medium staining and >175 for light staining. Fibres were counted and summed across entire gastrocnemius cross‐sections and subsequently analysed for group differences.

RNA sequencing and analysis

RNA was extracted from approximately 25 mg of gastrocnemius muscle as described above. A subset of samples (n = 2–3 mice per sex per group) were selected to roughly represent the average for each condition (male/female, tumour/PBS, PoWeR/sedentary) based on running distances, muscle and body mass. Library preparation was performed by the Oklahoma Medical Research Foundation (OMRF) Clinical Genomics Core and sequenced using an Illumina NovaSeq 6000 (150 bp paired‐end sequencing; Novogene Corp. Inc., Sacramento, CA, USA). Raw FASTQ files were initially uploaded to and processed in Partek Flow. Alignment was performed using STAR 2.7.8a, quantified to annotation model mm39, filtered for genes features with a maximum of <5 counts. Normalization and statistical comparisons were performed with DESeq2, excluding genes with a geometric average <1 over all samples prior to multiple test correction. Genes with a false discovery rate (Benjamini–Hochberg method) adjusted P‐value <0.05 were identified as differentially expressed genes (DEGs). Pathway analyses were performed on up‐ and downregulated DEGs in Enrichr (Xie et al., 2021) with the 2025 Gene Ontology (GO) database as the reference for our enrichment analysis. We used all protein‐coding genes detected in our RNA‐sequencing dataset as our background correction for the pathway analysis (Stokes et al., 2023). The top 10 gene sets within the Biological Process, Molecular Function, and Cellular Components from the GO database (by FDR) are reported.

Statistics

All data are presented as means ± SD. A two‐way ANOVA was conducted with tumour status (C26 vs. PBS) and intervention (PoWeR vs. SED) as factors to evaluate global effects on each dependent variable. When a significant interaction was detected, Tukey's post hoc test was applied to assess pairwise differences. An unpaired Student's t test was performed to compare the proportions of SDH stained fibres, as well as the AUC values for running distance and food consumption between sedentary tumour‐bearing mice (C26 SED) and exercised tumour‐bearing mice (C26 PoWeR). Additionally, a two‐way mixed‐design ANOVA followed by Tukey's post hoc test was conducted to evaluate both between‐subject and within‐subject effects on running distance and food consumption throughout the entire intervention period. It should be noted that food consumption could not be measured accurately in sedentary groups due to group housing. Therefore, only 4–6 individual values were available for statistical analysis. A one‐way ANOVA followed by Tukey's post hoc test was employed to compare muscle and organ weights (expressed as percentage differences from the healthy control group, PBS SED) among exercised non‐tumour‐bearing mice (PBS PoWeR), C26 SED and C26 PoWeR groups. Pearson's correlation coefficient (r) was used to examine associations between running distance (expressed as AUC) and muscle, fat, or tumour‐free body weights within exercised groups. Initial statistical analyses were performed within each biological sex (Appendix Figs A3, A4, A5, A6, A7, A8); however, few differences were observed between sexes and therefore all data were collapsed across sex for all analyses presented here. Statistical significance was defined a priori as P ≤ 0.05. All analyses were conducted using GraphPad Prism version 10.5.0 (GraphPad Software, Boston, MA, USA).

Results

Long‐term PoWeR training decreases tumour weight and preserves muscle and tumour‐free body weights in tumour‐bearing mice

All data were initially analysed within each sex (Appendix Figs A3, A4, A5, A6, A7, A8). Due to a lack of statistical differences, data were thereafter collapsed across sex for all further analyses presented here (PBS SED: n = 22, PBS PoWeR: n = 11, C26 SED: n = 18, C26 PoWeR: n = 11). Tumour‐bearing mice in the exercise group (C26 PoWeR) exhibited significantly lower tumour weight after ∼12 weeks of training (−38%) compared to sedentary tumour‐bearing mice (C26 SED) (P = 0.0018; Fig. 2A ). Tumour weights did not significantly correlate to muscle weights in exercised male and female mice (Appendix Fig. A1). Absolute tumour‐free body weight was not altered by PoWeR training (only main effect of tumour: P = 0.007; Fig. 2B ); however, a significant interaction between PoWeR training and tumour was found when tumour‐free body weight was expressed as a percentage change from either the initial body weight (IBW; at 8 weeks of age, prior to C26 implantation; P = 0.0156; Fig. 2C ) or from body weight at the day of C26 implantation (16 weeks of age; P = 0.0036; Fig. 2D ). Tukey's post hoc analysis revealed that sedentary tumour‐bearing mice had significantly lower tumour‐free body weight than all other groups: (i) expressed as percentage change from initial body weight: C26 SED (15%) vs. C26 PoWeR (26%; P < 0.0001), PBS PoWeR (30%; P < 0.0001) and PBS SED (27%; P < 0.0001); and (ii) expressed as percentage change from body weight at C26 implantation: C26 SED (−8%) vs. C26 PoWeR (−1%; P < 0.0001), PBS PoWeR (4%; P < 0.0001), PBS SED (2%; P < 0.0001) and C26 PoWeR (−1%) vs. PBS PoWeR (4%; P = 0.0235).

Figure 2. Twelve weeks of PoWeR training prevents cancer‐induced muscle wasting and cardiac atrophy, while attenuating splenomegaly and tumour growth in C26‐bearing mice.

Data for males (♂) and females (♀) are displayed separately in all panels but analysed together. A and B, tumour weight (g) (A) and tumour‐free body weight (g) (B) at the experimental endpoint. C and D, percentage change in tumour‐free body weight from baseline at 8 weeks of age (IBW: initial body weight) and at the time of tumour induction (16 weeks). E–G, muscle weights normalized to tibia length (mg/mm): ankle plantarflexors (soleus, plantaris, gastrocnemius), ankle dorsiflexors/knee extensors (tibialis anterior [TA], extensor digitorum longus [EDL], quadriceps) and heart. H, organ weights normalized to tibia length (mg/mm): liver, spleen and fat. I and J, heatmaps showing percentage differences in normalized skeletal muscle (I) and organ weights (J) among PBS PoWeR (n = 11), C26 SED (n = 18), and C26 PoWeR (n = 11) groups relative to healthy controls (PBS SED; n = 22). Red asterisks indicate significant tumour main effects (C26 vs. PBS, P < 0.05), and blue asterisks indicate significant exercise main effects (PoWeR vs. SED, P < 0.05), determined by two‐way ANOVA. Coloured outlines denote significant post hoc Tukey's tests (P < 0.05): light blue (C26 SED vs. C26 PoWeR: soleus, plantaris, TA, heart in I; spleen in J), orange (PBS PoWeR vs. C26 SED: soleus, TA, EDL, heart in I; fat, spleen in J) and light green (PBS PoWeR vs. C26 PoWeR: fat, spleen in J). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

To account for any inherent differences in body size, all tissue weights were normalized by tibia length as a surrogate marker for overall body size. An interaction between PoWeR training and tumour was also observed in normalized hindlimb muscle weights (Fig. 2E and F ), with the exception of the extensor digitorum longus (EDL), where only a tumour main effect was found (P = 0.0003). Specifically, in the soleus, C26 SED mice had significantly lower normalized weights compared to all other groups (interaction: P = 0.0011; C26 SED vs. C26 PoWeR, −31%, P = 0.0004; vs. PBS PoWeR, −22%, P = 0.0222; vs. PBS SED, −26%, P = 0.0005). Normalized weights of plantaris, gastrocnemius and quadriceps were all lower in C26 SED compared to PBS SED (interaction P = 0.0086, 0.0473, 0.0052, respectively; pairwise comparisons: −23%, −17%, −22% and P = 0.0024, 0.0088, 0.0001, respectively). Tibialis anterior (TA) was also lower in C26 SED compared to both PBS groups (interaction P = 0.0089; C26 SED vs. PBS PoWeR, −12%, P = 0.0206; vs. PBS SED, −18%, P < 0.0001). For non‐muscle tissues, a significant tumour effect (P = 0.0029) and PoWeR effect (P = 0.0241), but no interaction, was observed in heart weight (Fig. 2G ). Gonadal fat was reduced by tumour (P = 0.0003; Fig. 2H ). In spleen weight, a strong interaction between tumour and PoWeR (P < 0.0001) was observed. C26 SED mice exhibited markedly elevated spleen weights compared to all groups (C26 PoWeR: 33%, PBS PoWeR: 63%, PBS SED: 59%; all P < 0.0001). Furthermore, C26 PoWeR mice had higher spleen weights than both PBS groups (PBS PoWeR: 44%, PBS SED: 38%; P < 0.0001 for both comparisons) (Fig. 2H ). Comparable results were obtained with raw (non‐normalized) values of muscle and organ weights (Appendix Fig. A2).

When normalized weights were expressed as a percentage difference relative to PBS SED controls, PoWeR training attenuated cancer‐induced muscle wasting and splenomegaly. This was evident in significant differences between C26 SED and C26 PoWeR mice (soleus: −20% vs. 4%, P = 0.0002; plantaris: −19% vs. −3%, P = 0.0363; TA: −15% vs. −7%, P = 0.0347; heart: −15% vs. −3%, P = 0.0258; spleen: 143% vs. 62%, P < 0.0001) (Fig. 2I and J ). Additional significant differences were observed between PBS PoWeR and C26 SED groups (soleus: −3% vs. −20%, P = 0.0081; TA: −5% vs. −15%, P = 0.0065; EDL: −4% vs. −17%, P = 0.017; heart: 3% vs. −15%, P = 0.0007; fat: −35% vs. −58%, P = 0.0214; spleen: −9% vs. 143%, P < 0.0001) as well as between PBS PoWeR and C26 PoWeR groups (fat: −35% vs. −60%, P = 0.0252; spleen: −9% vs. 62%, P < 0.0001) (Fig. 2I and J ).

Overall distance during PoWeR training is maintained regardless of tumour status, while food consumption is increased with exercise

When individual time points were analysed by two‐way mixed‐design ANOVA, running distance differed between 2 and 7 weeks in PBS PoWeR (P = 0.0322) and between 2 weeks and 10 weeks, 11 weeks and endpoint in C26 PoWeR (P = 0.0439, 0.0274, 0.0249, respectively) (Fig. 3A ). However, no difference was observed in overall running distance (overall, pre‐ and post‐tumour inoculation) between exercised tumour‐bearing (n = 11) and non‐tumour‐bearing mice (n = 11), as assessed by AUC (Fig. 3B–D ).

Figure 3. Running performance was unaffected by tumour status, while food intake increased with exercise.

A–D, weekly (A) and cumulative (B) running performance (area under the curve, AUC) throughout the intervention, during the preconditioning period (weeks 1 to 8; prior to tumour inoculation) (C), and following tumour inoculation (week 9 to endpoint) (D). (E and F, weekly (E) and cumulative (F) food consumption (AUC) across the intervention. G, grip strength normalized to tibia length. Blue asterisk indicates significant main effects of exercise (PoWeR vs. SED, P < 0.05), and purple asterisk indicates significant exercise × tumour interactions (P < 0.05), determined by two‐way ANOVA. PBS SED: n = 22, PBS PoWeR: n = 11, C26 SED: n = 18, C26 PoWeR: n = 11. Data for males (♂) and females (♀) are displayed separately in panels B, C, D, F and G but analysed together.

Significant differences were found in food consumption across groups (PBS SED: n = 22, PBS PoWeR: n = 11, C26 SED: n = 18, C26 PoWeR: n = 11) throughout the intervention period (weeks 1–11), including PBS SED vs. PBS PoWeR (P < 0.0001–0.05), PBS SED vs. C26 PoWeR (P < 0.0001–0.05), PBS PoWeR vs. C26 SED (P < 0.0001–0.001) and C26 SED vs. C26 PoWeR (P < 0.0001–0.001) (Fig. 3E ). At the endpoint, differences remained only between PBS SED vs. PBS PoWeR (P = 0.0036) and PBS PoWeR vs. C26 SED (P = 0.0146). A main effect of PoWeR training was found for overall food consumption (P < 0.0001), assessed by AUC, with no significant differences within PoWeR or sedentary groups (Fig. 3F ). Food consumption in sedentary groups was estimated as the cage‐average per mouse due to group housing.

A significant interaction between PoWeR training and tumour status was observed for grip strength normalized to tibia length (P = 0.049); however, post hoc analysis revealed no significant between‐group differences in grip strength (Fig. 3G ).

Long‐term PoWeR training suppresses markers of protein degradation without altering protein synthesis, while restoring cancer‐induced dysregulation of mitophagy

Protein fractional synthetic rate did not differ significantly between groups (Fig. 4A ). To further examine canonical regulation during CC we performed a series of a priori assessments by RT‐PCR to determine impacts on markers of protein turnover and mitochondrial quality control (PBS SED: n = 18–22, PBS PoWeR: n = 10–11, C26 SED: n = 16–18, C26 PoWeR: n = 9–11). These specific markers and the analysis strategy were chosen to be consistent with our prior work and allow for direct comparisons (Cabrera et al., 2023; Tsitkanou et al., 2024, 2026). A significant interaction between PoWeR training and tumour status was observed for the anabolic suppressor Redd1 (interaction: P = 0.0292; C26 SED vs. PBS SED: P < 0.0001; C26 SED vs. PBS PoWeR: P = 0.0009) (Fig. 4B ), as well as for the catabolic markers Murf1, Atrogin, Ubc and Gadd45a (interaction: P < 0.0001–0.01; C26 SED vs. C26 PoWeR: P < 0.0001–0.01; C26 SED vs. PBS SED: P < 0.0001; C26 SED vs. PBS PoWeR: P < 0.0001–0.01) (Fig. 4C ). Similarly, an interaction was found for the mitochondrial fission marker Bnip3 (interaction: P = 0.0007; C26 SED vs. C26 PoWeR: P = 0.0002; C26 SED vs. PBS SED: P < 0.0001; C26 SED vs. PBS PoWeR: P < 0.0001) (Fig. 4D ). In all cases, mRNA expression of these markers was highest in the C26 SED group compared with all other groups. Unexpectedly, a significant interaction between PoWeR training and tumour status was also detected for the mitochondrial biogenesis factor Pgc1a (P = 0.0235), with expression being higher in C26 SED compared to PBS SED (P = 0.0064) (Fig. 4D ).

Figure 4. PoWeR training attenuated protein degradation markers while leaving protein synthesis unchanged, and reversed cancer‐associated alterations in mitophagy.

A, protein fractional synthetic rate (%/h) in gastrocnemius. B–D, gene expression (normalized to 18S) of anabolic repressors (Deptor, Redd1), catabolic markers (Murf1, Atrogin, Ubc, Gadd45a) and mitochondrial markers (Opa1, Bnip3, Pgc1a) in gastrocnemius. Red asterisks denote significant tumour main effects (C26 vs. PBS, P < 0.05), and blue asterisks denote significant exercise main effects (PoWeR vs. SED, P < 0.05), determined by two‐way ANOVA. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. PBS SED: n = 18–22, PBS PoWeR: n = 10–11, C26 SED: n = 16–18, C26 PoWeR: n = 9–11. Data for males (♂) and females (♀) are displayed separately in all panels but analysed together. AU: arbitrary units.

Long‐term PoWeR training promotes a faster‐to‐slower muscle fibre‐type transition independent of tumour status

A main effect of PoWeR training was observed in the average fibre CSA of the gastrocnemius, with exercised groups (PBS PoWeR: n = 9 and C26 PoWeR: n = 9) having smaller fibre size than sedentary groups (PBS SED: n = 13 and C26 SED: n = 12; P = 0.0484) (Fig. 5A ). A tumour main effect was also found in the CSA of IIX+B fibres, as tumour‐bearing mice exhibited smaller CSA compared to controls (P = 0.0218) (Fig. 5A ). PoWeR significantly influenced fibre‐type distribution, showing a greater proportion of type IIA fibres (P < 0.0001) with a lower proportion of type IIX+B fibres (P = 0.0008). This indicates a shift from much larger fast‐twitch IIX+B fibres toward smaller and slower‐twitch IIA fibres in response to exercise (Fig. 5B and C ).

Figure 5. PoWeR training promoted a fast‐to‐slow muscle fibre‐type transition independent of tumour status while reducing average fibre size in the gastrocnemius.

A, fibre cross‐sectional area (CSA, µm²) presented as the average of all fibre types and separately for type I, IIA, and IIX+B fibres. B, muscle fibre‐type composition in the gastrocnemius. C, representative immunofluorescence images (×20 magnification): MyHC I fibres (pink), MyHC IIA fibres (green), MyHC IIX+B fibres (unstained/black), and sarcolemma (red). Scale bar is 100 µm. Red asterisks indicate significant tumour main effects (C26 vs. PBS, P < 0.05), and blue asterisks indicate significant exercise main effects (PoWeR vs. SED, P < 0.05), determined by two‐way ANOVA. PBS SED: n = 13, PBS PoWeR: n = 9, C26 SED: n = 12, C26 PoWeR: n = 9. Data for males (♂) and females (♀) are displayed separately in panel A but analysed together.

Long‐term PoWeR training induces a trend toward improved muscle oxidative capacity in tumour‐bearing mice

Proportions of SDH‐positive fibres classified as light, medium or dark staining in the C26 SED gastrocnemius (n = 13) were 0.17 ± 0.07, 0.61 ± 0.12 and 0.23 ± 0.09, respectively (Fig. 6A and  C ). In the C26 PoWeR gastrocnemius muscle (n = 6), the proportions of light‐, medium‐ and dark‐stained fibres were 0.27 ± 0.15, 0.77 ± 0.10 and 0.24 ± 0.09, respectively (Fig. 6B and  C ). The proportion of medium‐stained fibres in the C26 PoWeR group showed a trend toward a 26% increase compared with C26 SED (0.77 ± 0.10 vs. 0.61 ± 0.12, P = 0.060).

Figure 6. PoWeR training did not significantly alter muscle oxidative capacity in tumour‐bearing mice.

A and B, succinate dehydrogenase (SDH)‐stained gastrocnemius sections from sedentary (C26 SED; n = 13) (A) and exercised (C26 PoWeR; n = 6) (B) tumour‐bearing mice following 11.5 weeks of PoWeR training. C, SDH‐positive fibres were categorized by staining intensity (light, medium, and dark). Exercise resulted in a non‐significant increase in the proportion of medium‐stained fibres (P = 0.06), suggesting a trend toward improved oxidative capacity. Data represent combined analysis of males and females.

Total running distance in exercise groups correlates negatively with tumour‐free body weight, gonadal fat and muscle mass

Total running distance in exercised mice (n = 22) was negatively correlated with tumour‐free body weight at endpoint (r = –0.4593, P = 0.0315) (Fig. 7A ) and with gonadal fat (r = –0.5383, P = 0.0098) (Fig. 7B ). Running distance was also negatively correlated with the weights of plantaris (r = −0.4292, P = 0.0462) (Fig. 7C ), gastrocnemius (r = −0.6189, P = 0.0021) (Fig. 7D ) and quadriceps (r = −0.47, P = 0.0273) (Fig. 7E ). Pearson's r and P‐values for correlations between running distance and muscle, organ and body weights, analysed in both exercised groups combined and separately, are provided in Appendix Table A1. Pearson's correlation analyses were conducted using organ and muscle weights normalized to tibia length, as well as running distance expressed as the AUC.

Figure 7. Total running distance, expressed as area under the curve (AUC), showed inverse correlations with tumour‐free body weight (g) (A), gonadal fat (mg/mm) (B), and muscle mass (mg/mm) of the plantaris (C), gastrocnemius (D), and quadriceps (E).

Stars represent values from C26 tumour‐bearing mice (n = 11), while circles represent values from non–tumour‐bearing mice (n = 11). Values shown represent Pearson's correlation coefficient (r) and the corresponding statistical significance (P).

RNA‐sequencing revealed that PoWeR training corrected a deleterious metabolism‐oriented transcriptional response to CC

We previously reported transcriptomic alterations within several models of CC in male and female mice (Blackwell et al., 2018; Morena et al., 2024). To globally evaluate how long‐term exercise training (PoWeR) influences the transcriptome during cachexia, we performed RNA sequencing on the gastrocnemius muscle, a primary hindlimb locomotive muscle during PoWeR which is commonly affected during cachexia, on a subset of male and female mice from each condition (n = 2–3 per sex per group). Principal component analysis demonstrates clear discrimination between the experimental conditions in our study (Fig. 8A ). To first consider the effects of only PoWeR, we compared healthy PoWeR‐trained and sedentary mice (PBS PoWeR vs. PBS SED). Forty‐two differentially expressed genes (DEGs; adj. P < 0.05; Fig. 8B and  B′ ) were observed between healthy PoWeR‐trained and sedentary mice (PBS PoWeR and PBS SED). The top three upregulated genes by fold‐change were myosin light chain 6b (Myl6b, adj. P = 0.0112), erythroferrone (Erfe, adj. P = 0.0479) and SH3 domain binding kinase 2 (Sbk2 adj. P = 0.00464, Fig. 8B′ ). Three of the top downregulated genes were insulin‐like growth factor 1 (Igf1, adj. P = 0.0177), macrophage immunometabolism regulator (Macir, adj. P = 0.0177) and phosphorylase kinase alpha 1 (Phka, adj. P = 0.0099, Fig. 8B′ ). Between tumour bearing sedentary mice (C26 SED) and healthy sedentary mice (PBS SED), there were 2740 DEGs (Fig. 8C ). We performed GO enrichment analyses (biological processes, molecular function, cellular component). With C26 SED, we observed upregulation of pathways related to cytoplasmic translation, ribosome, ribonucleoprotein complex biogenesis and ribosome biogenesis (Fig. 8D ). Downregulated GO pathways were overwhelmingly related to mitochondrial regulation and oxidative bioenergetics (Fig. 8E ). To assess how PoWeR may have impacted the cachectic transcriptome we compared C26 PoWeR to C26 SED. Between C26 tumour bearing SED and PoWeR‐trained mice (C26 PoWeR), there were 963 DEGs (Fig. 8F ). Upregulated GO functional annotations in C26 PoWeR versus C26 SED were related to aerobic respiration and mitochondrial processes, and downregulated GO pathways were related to the extracellular matrix, secretory granules and endocytosis, and regulation of macrophage activation (Fig. 8G and  H , respectively). The top 10 GO terms downregulated in C26 SED were also the top 10 terms upregulated in C26 PoWeR (Fig. 8E and  G ). Therefore, we evaluated the specific genes within those pathways, providing 131 genes which appear to be ‘corrected’ by PoWeR training. Expression of these genes is shown in a Venn diagram and heatmap (Fig. 8H and  I , respectively). Specific genes following this pattern include citrate synthase (Cs), succinate dehyodrogenase subunit B (Sdhb) and numerous subunits of the respiratory electron transport chain (cytochrome oxidase, Cox‐; mitoribosomal genes, Mrpl‐; NADH:ubiquinone oxidoreductases, Ndufa‐, Ndufb‐).

Figure 8. PoWeR training counteracts tumour‐induced mitochondrial and metabolic transcriptomic alterations in skeletal muscle.

A, principal component analysis (PCA) plot of samples. B, volcano plot of differentially expressed genes between PBS PoWeR and PBS SED. B′, heatmap showing z‐scores of all differentially expressed genes between PBS PoWeR and PBS SED. C, volcano plot of differentially expressed genes between C26 SED and PBS SED. D, top 10 GO functional annotations from upregulated genes in C26 SED vs. PBS SED. E, top 10 GO functional annotations from downregulated genes in C26 SED vs. PBS SED. F, volcano plot of differentially expressed genes between C26 PoWeR and C26 SED. G, top 10 GO functional annotations from upregulated genes in C26 PoWeR vs. C26 SED. H, top 10 GO functional annotations from downregulated genes in C26 PoWeR vs. C26 SED. I, comparison of mitochondrial genes from the top 10 up/down GO pathways common to C26 SED vs. PBS SED and C26 PoWeR vs. C26 SED. J, heatmap showing z‐scores of all 131 overlapping mitochondrial‐related differentially expressed genes (adj. P < 0.05). PBS SED: n = 6, PBS PoWeR: n = 6, C26 SED: n = 5, C26 PoWeR: n = 5.

Discussion

Exercise training has been proposed as a therapeutic approach to attenuate cancer‐induced muscle wasting (Lira et al., 2014; Tsitkanou et al., 2022), with a growing body of evidence supporting concurrent training (resistance and endurance exercise) as the optimal modality to mitigate CC. Several studies have shown that concurrent training, initiated either after tumour induction (Ballaro et al., 2019; Collao et al., 2023) or before and maintained throughout disease progression (Ranjbar et al., 2019; Wood et al., 2022), may preserve muscle mass. Here, we demonstrate that exercise pre‐conditioning maintained through disease progression preserves tumour‐free body and muscle mass (soleus, plantaris, TA), reduces tumour burden, induces a fibre‐type transition toward a more oxidative phenotype, and counteracts deleterious transcriptional responses associated with CC. These adaptations were accompanied by reduced expression of genes related to protein degradation and mitophagy. The current findings contrast with our previous report showing that short‐term PoWeR in young mice did not result in larger muscles in both forelimbs and hindlimbs, irrespective of tumour presence (Tsitkanou et al., 2026). Perhaps extending the training duration from 3.5 to 11.5 weeks (largely performed prior to tumour implantation) and increasing resistance from 4 g to 6 g led to more robust protective effects and allowed sufficient time for cellular adaptations to emerge, which may not have occurred with shorter interventions in young developing mice. The 8 weeks of PoWeR training prior to cancer onset likely offered substantial pre‐conditioning benefits that contributed to these protections. This interpretation is supported by findings from Hiroux et al. (2021) showing that short‐term (˜2.5 weeks) voluntary wheel running without resistance failed to preserve muscle mass in C26‐bearing mice. Importantly, PoWeR training also protected against cancer‐induced cardiac atrophy, extending our earlier findings (Tsitkanou et al., 2026) and aligning with previous evidence of PoWeR benefits for the heart (Dungan et al., 2019).

A novel observation in the present study is that exercise reduced tumour weight, a result not consistently reported with other concurrent training protocols (Ballaro et al., 2019; Ranjbar et al., 2019; Tsitkanou et al., 2026; Wood et al., 2022). Aerobic exercise appears particularly relevant for tumour control, as in prior works treadmill running reduced large tumour incidence in Apc^Min/+ mice by 20% (Puppa et al., 2012), while resistance training modestly (though non‐significantly) increased tumour weight in C26 mice (Khamoui et al., 2016). Emerging evidence indicates that aerobic exercise (voluntary wheel running) may slow tumour progression by redirecting systemic glucose utilization from the tumour microenvironment to metabolically demanding tissues such as skeletal and cardiac muscle (Leitner et al., 2025). In our previous work, short‐term PoWeR did not affect tumour mass in either males or females (Tsitkanou et al., 2026). Thus, suppression of tumour‐promoting mechanisms may require sustained training durations to manifest (Bettariga et al., 2024). The lack of a significant relationship between muscle mass and tumour size suggests muscle mass preservation was not just a consequence of lower tumour burden from exercise (Appendix Fig. A1). Exercise protects skeletal muscle irrespective of tumour dynamics, although the underlying mechanism was not investigated in the present study.

Throughout the 11.5‐week exercise intervention, C26 tumour‐bearing mice maintained consistent running performance, with no significant differences in distance covered compared to controls. This resilience contrasts with shorter voluntary wheel running studies (2.5–3.5 weeks post‐tumour induction) that reported impaired running performance accompanied by decreased food consumption (Pigna et al., 2016; Tsitkanou et al., 2026). Our findings, together with Collao et al. (2023), suggest that longer training durations allow for the establishment of stable muscle adaptations that sustain exercise capacity even during CC symptomatic stages, provided food intake remains intact. This contrasts with Apc^Min/+ mice, which exhibit a decline in endurance capacity during late‐stage cachexia despite early initiation and prolonged training (21 weeks), likely driven by anaemia (Baltgalvis et al., 2010). However, the severity of the condition at endpoint in Apc^Min/+ mice stands in stark contrast to their maintained ability to remain active.

In addition, long‐term PoWeR training suppressed canonical markers of protein degradation (Murf1, Atrogin‐1, Ubc, Gadd45a) and the anabolic repressor Redd1, while protein synthesis rates remained unaltered across conditions. These effects diverged from our prior short‐term PoWeR study (Tsitkanou et al., 2026), in which PoWeR failed to mitigate the induction of pro‐catabolic markers in tumour‐bearing mice. Collectively, this suggests that short‐term exercise at symptomatic stages is insufficient to counteract tumour‐induced muscle disruptions if undertaken without prior conditioning, and that controlled workloads may be required when muscle integrity is already compromised.

PoWeR training also displayed a more exercise‐conditioned muscle phenotype. These adaptations included a shift from fast‐glycolytic type IIX and IIB fibres toward fast‐oxidative type IIA fibres in the gastrocnemius, which was accompanied by lower mean fibre CSA relative to sedentary animals. A fibre type transition and relatively smaller muscle fibres are both classical adaptations to endurance exercise in mice and humans (Guderley et al., 2008; Yan et al., 2011). Our data agree with our previous work wherein PoWeR initiated in young growing mice did not elicit hypertrophy (Tsitkanou et al., 2026), whereas in more mature adult mice (>4 months of age), hypertrophy is characteristic (Dungan et al., 2019; Murach et al., 2020). There was an inverse relationship between muscle mass (plantaris, gastrocnemius, quadriceps) and running distance, consistent with our earlier findings (Tsitkanou et al., 2026). These data suggest the need for exercise load/volume management and personalized exercise prescription to facilitate hypertrophy and combat cachexia. PoWeR training suppressed the cachexia‐induced increase in expression of Bnip3, a mitophagy marker upregulated in C26 mice. While Bnip3 suppression attenuates fat loss and splenomegaly in LLC models, it is insufficient alone to prevent muscle loss (Morena et al., 2025).

Our transcriptomic analysis suggests an improvement in mitochondrial integrity and function with PoWeR. In C26‐bearing mice, we observed downregulation of pathways related to mitochondrial membranes, aerobic respiration and oxidative phosphorylation. These are processes we have consistently shown to be dysregulated during CC and even to precede muscle wasting (Blackwell et al., 2018; Brown et al., 2017; Morena et al., 2024). PoWeR appeared to reverse this signature, as all 10 of the top downregulated pathways in C26 mice were upregulated in tumour‐bearing mice that underwent PoWeR training. This molecular signature from PoWeR highlights the potent capacity of exercise to preserve mitochondrial transcriptome integrity in the context of CC. Although we did not further assess these effects with functional measures, our prior work with PoWeR in aged mice shows a profound remodelling of the muscle proteome that reflects the PoWeR RNA‐seq data here, specifically with respect to upregulated mitoribosomal and mitochondrial proteins (Chambers et al., 2025). In addition, SDH staining of the gastrocnemius suggested a trend toward improved oxidative capacity in exercised tumour‐bearing mice. Overall, these data suggest that PoWeR is able to protect against key cellular features of cachexia, which may be critical to improve outcomes.

We acknowledge several limitations of the present study. First, the sample size within each exercised group, when stratified by biological sex, was insufficient to permit adequately powered statistical comparisons. Although sexual dimorphism has been well established in C26 mice (Cabrera et al., 2023; Halle et al., 2026; Tsitkanou et al., 2024) and, more broadly, in CC in both clinical (Zhong & Zimmers, 2020) and preclinical settings (Morena et al., 2024), we combined male and female mice in the primary analyses. This decision was based on our previous study using the PoWeR model (Tsitkanou et al., 2026) and the findings of the present study, which did not reveal meaningful sex‐specific differences in response to the exercise intervention. While clustering in phenotypic and running distance data was observed between sexes, these patterns are expected given established sex differences with males exhibiting greater body size reflected in body and tissue weights and muscle fibre CSA, whereas females typically demonstrate higher voluntary running distances (Tsitkanou et al., 2026). Despite these differences in absolute values, the direction and overall trends in phenotypic outcomes in response to cancer and PoWeR training were consistent between sexes, with exercise preserving muscle mass and suppressing tumour growth in males and females. Consistent with both statistical and qualitative assessments, we did not detect meaningful sex‐dependent differences in response to exercise intervention. As sex did not affect response to our interventions, data are presented collapsed across sexes, as pooling enhances interpretability without compromising the detection of biologically relevant effects. For transparency, all analyses and data stratified by biological sex are now provided in the Appendix (Figs A3, A4, A5, A6, A7, A8) and FigShare (‘Data availability statement’). In addition, another limitation of this study is that functional assessments were limited to grip strength measurements. The absence of more comprehensive evaluations of functional performance, such as in vivo muscle contractility and aerobic capacity, limits our ability to fully interpret changes in muscle contractile performance. Inclusion of these measures in future studies would allow a more comprehensive consideration of muscle health following exercise. Furthermore, while cachexia is not typically influenced by dietary intervention (Fearon et al., 2011), increased food intake in the exercise groups may have contributed to the attenuation of cachexia.

We also acknowledge that although we cannot entirely exclude the possibility that exercise continuing uninterrupted through tumour inoculation may have inadvertently influenced tumour development, this appears unlikely. In our recent study (Tsitkanou et al., 2026), animals began wheel acclimation 1 week prior to tumour inoculation and PoWeR training commenced immediately after inoculation, under conditions similar to those used in the present study. In that study, we observed no difference in tumour size between sedentary and PoWeR‐exercised animals. Taken together, these findings suggest that the exercise preconditioning component of the present study may have contributed to the reduced tumour volume observed. Similar effects have also been reported in a recent study using voluntary exercise as a preconditioning intervention (Leitner et al., 2025). However, this interpretation remains speculative and further studies are required to clarify this potential mechanism. Although only endpoint tumour weights were assessed in the present study, the inclusion of longitudinal tumour volume measurements in addition to terminal weights would strengthen the evaluation of tumour growth suppression. Incorporating such measurements would be valuable in future studies. Finally, the relationship between exercise volume and body weight in tumour‐bearing mice is complex. Although exercised mice maintained higher tumour‐free body weight, some ran shorter distances. This likely reflects the concurrent nature of PoWeR training, in which the resistance component supports muscle preservation, while the endurance component influences total body weight and running capacity. Consequently, body weight and exercise volume do not always correlate directly in PoWeR training.

In summary, our findings demonstrate that long‐term PoWeR training, initiated prior to the onset of cachexia, provides protection against cancer‐induced muscle and body mass loss, preserves cardiac tissue, and uniquely reduces tumour burden in C26‐bearing mice. These benefits are driven by exercise‐induced molecular and cellular adaptations, including suppression of protein degradation pathways, promotion of oxidative fibre‐type remodelling and improvements in the mitochondrial transcriptome. Importantly, the preservation of skeletal muscle occurred independently of tumour growth attenuation, indicating that exercise exerts independent benefits on muscle health in addition to its potential to slow tumour progression. These findings underscore the preventive and therapeutic potential of concurrent endurance and resistance exercise as a non‐pharmacological intervention to combat CC. Moreover, our findings emphasize the value of exercise pre‐conditioning as a critical determinant of its effectiveness.

Competing interests

No conflicts of interest, financial or otherwise, are declared by the authors.

Author contributions

The manuscript was drafted by S.T. and P.J.K. Experimental design was conceived by N.P.G., S.T., K.A.M. and P.J.K. Animal experiments were performed by S.T. and P.J.K. Analyses were performed by S.T., P.J.K., A.R.C., N.S., C.P., R.M., Z.B.M., M.G., E.R.S., F.M., and Y.W. S.T., N.P.G., K.A.M. and T.W. supervised and obtained funding for this work. All authors contributed intellectually and provided feedback. All authors have read and approved the final version of this manuscript and agree to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. All persons designated as authors qualify for authorship, and all those who qualify for authorship are listed.

Funding

This study was funded by the Women's Giving Circle Grant (University of Arkansas, 2023–2025) (S.T.), the National Institutes of Health: National Institute of Arthritis and Musculoskeletal and Skin Diseases R01AR075794‐01A1/AR/NIAMS (N.P.G.), the National Institute of General Medical Sciences award number P20GM125503 (N.P.G.).

Supporting information

Peer Review History

Biographies

Stavroula Tsitkanou is a Postdoctoral Research Fellow in the Department of Neurology at Beth Israel Deaconess Medical Centre and Harvard Medical School. Prior to this role, she was a Postdoctoral Research Fellow at the Cachexia Research Laboratory at University of Arkansas. She received her Ph.D. from Deakin University (Australia), and her research focuses on muscle biology and exercise physiology in the context of compromised skeletal muscle conditions.

Pieter J. Koopmans recently earned his Ph.D. in the Molecular Muscle Mass Regulation (M3R) Laboratory at the University of Arkansas under the supervision of Dr. Kevin A. Murach. He received his undergraduate degree in Exercise and Sport Science from University of Wisconsin – La Crosse in La Crosse, Wisconsin. He then went on to complete a master's degree in Exercise Science at Appalachian State University in Boone, North Carolina. His research focus includes the myonuclear epigenetic and molecular regulation of skeletal muscle ageing and exercise adaptation, and he has a particular interest in developing strategies to mitigate muscle wasting during acute and chronic illness.

Table A1.

Correlations between total running distance, expressed as area under the curve (AUC), and tumour‐free body weight (TFBW; raw values and percentage change from initial body weight), tumour weight and tissue weights normalized to tibia length

	Correlation with running distance (AUC)
	r	P
Both C26 and PBS mice (n = 22)
TFBW	−0.4593	0.0315
Body weight % change from initial body weight (7 weeks old)	0.4039	0.0623
Soleus	−0.4144	0.0552
Plantaris	−0.4292	0.0462
Gastrocnemius	−0.6189	0.0021
TA	−0.2352	0.2921
EDL	−0.0542	0.8107
Quadriceps	−0.47	0.0273
Heart	−0.2378	0.2865
Fat	−0.5383	0.0098
Liver	−0.2958	0.1814
Spleen	−0.07592	0.737
Tumour	−0.121	0.723
Only C26 mice (n = 11)
TFBW	−0.1765	0.6038
Body weight % change from initial body weight (7 weeks old)	0.3557	0.283
Soleus	−0.1266	0.7108
Plantaris	−0.3185	0.3397
Gastrocnemius	−0.4536	0.1611
TA	−0.2812	0.4023
EDL	−0.2146	0.5264
Quadriceps	−0.2471	0.4638
Heart	−0.1504	0.6589
Fat	−0.5806	0.0611
Liver	−0.009179	0.9786
Spleen	0.1642	0.6295
Tumour	−0.121	0.723
Only PBS mice (n = 11)
TFBW	−0.7897	0.0038
Body weight % change from initial body weight (7 weeks old)	0.4138	0.2058
Soleus	−0.7648	0.0061
Plantaris	−0.5286	0.0946
Gastrocnemius	−0.8186	0.0021
TA	−0.2939	0.3803
EDL	−0.04908	0.886
Quadriceps	−0.6597	0.0272
Heart	−0.4167	0.2023
Fat	−0.8255	0.0018
Liver	−0.5192	0.1017
Spleen	−0.0926	0.7866

Pearson's r analyses were performed across all mice (PBS and C26 combined, n = 22) as well as within PBS (n = 11) and C26 (n = 11) groups separately. P‐values shown in bold represent statistical significance.

Figure A1. (A) Soleus, (B) plantaris, and (C) gastrocnemius muscle weights were not significantly correlated with tumor weight, suggesting that exercise‐induced preservation of muscle mass is independent of tumor burden.

Pearson's r correlations between tumour weight and tibia length‐ normalized muscle weights (mg/mm) of ankle plantarflexors (soleus, plantaris, gastrocnemius) in exercised mice

Figure A2. Raw muscle weights (mg) of ankle plantarflexors (soleus, plantaris, gastrocnemius), ankle dorsiflexors/knee extensors (tibialis anterior [TA], extensor digitorum longus [EDL], quadriceps) and heart (A–C), as well as raw organ weights of liver, spleen and fat (D).

Red asterisks denote significant tumour main effects (C26 vs. PBS, P < 0.05), and blue asterisks denote significant exercise main effects (PoWeR vs. SED, P < 0.05), determined by two‐way ANOVA. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Figure A3. PoWeR training prevents cancer‐induced muscle wasting and cardiac atrophy, as well as mitigating cancer‐induced splenomegaly and tumour growth in males.

A and B, tumour‐free body weight (g) (A) and tumour weight (mg) (B) at the experimental endpoint. C and D, percentage difference of tumour‐free body weight at the experimental endpoint from the initial body weight at 8 weeks old (C) and at the day of tumour induction (16 weeks old) (D). E and F, muscle weights normalized by initial body weight (IBW) (mg/g) of soleus, plantaris, gastrocnemius, tibialis anterior (TA), extensor digitorum longus (EDL), quadriceps and heart (E), as well as organ weights normalized by IBW (mg/g) of liver, spleen and fat (F) at the experimental endpoint (25 days after the inoculation of C26 cells) in males. G and H, heatmaps of the percentage difference of C26 SED and C26 PoWeR from the healthy control group (PBS SED) in skeletal muscle tissues (G) and organs (H) in males. Red asterisks symbolize a significant (P < 0.05) tumour main effect (C26 vs. PBS), while blue asterisks symbolize a significant (P < 0.05) PoWeR exercise main effect (PoWeR vs. SED) after performing 2‐way ANOVA analysis. Red squares outlining soleus, plantaris and heart in panel G, as well as spleen in panel H, symbolize significant differences (P < 0.05) between C26 SED vs. C26 PoWeR by unpaired t test. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Figure A4. Similarly with males, PoWeR training mitigates and sometimes prevents cancer‐induced muscle wasting and cardiac atrophy, as well as mitigating cancer‐induced splenomegaly and tumour growth in females.

A and B, tumour‐free body weight (g) (A) and tumour weight (mg) (B) at the experimental endpoint. C and D, percentage difference of tumour‐free body weight at the experimental endpoint from the initial body weight at 8 weeks old (C) and at the day of tumour induction (16 weeks old) (D). E and F, muscle weights normalized by initial body weight (IBW) (mg/g) of soleus, plantaris, gastrocnemius, tibialis anterior (TA), extensor digitorum longus (EDL), quadriceps and heart (E), as well as organ weights normalized by IBW (mg/g) of liver, spleen and fat (F) at the experimental endpoint (25 days after the inoculation of C26 cells) in females. G and H, heatmaps of the percentage (%) difference of C26 SED and C26 PoWeR from the healthy control group (PBS SED) in skeletal muscle tissues (G) and organs (H) in females. Red asterisks symbolize a significant (P < 0.05) tumour main effect (C26 vs. PBS), while blue asterisks symbolize a significant (P < 0.05) PoWeR exercise main effect (PoWeR vs. SED) after performing 2‐way ANOVA analysis. Red squares outlining soleus, plantaris, quadriceps and heart in panel G, as well as spleen in panel H, symbolize significant differences (P < 0.05) between C26 SED vs. C26 PoWeR by unpaired t test. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Figure A5. Tumor burden does not alter running distance or food intake in exercised male and female mice.

Running distance and food consumption did not change during cancer cachexia (CC) progression in exercised male (A and C) and female mice (E and G). Area under the curve (AUC) in running distance and food consumption was also calculated and did not differ between exercised non‐tumour‐bearing mice and exercised tumour‐bearing mice in both males (B and D) and females (F and H).

Figure A6. Exercise suppressed muscle catabolic markers in tumour‐bearing male and female mice, without altering protein synthesis rates.

Female mice exhibited preservation of mitochondrial markers against tumour‐induced degeneration. Protein synthesis rate is shown for males (A) and females (E). Gene expression of anabolic repressors, catabolic markers, and mitochondrial markers is presented for males (B, C and D, respectively) and females (F, G and H, respectively). Red asterisks indicate a significant tumour main effect (C26 vs. PBS; P < 0.05), whereas blue asterisks indicate a significant PoWeR exercise main effect (PoWeR vs. SED; P < 0.05), as determined by two‐way ANOVA. #Significant Tumour and PoWeR interaction. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Figure A7. An increase in the percentage of type IIA fibres was observed after 12 weeks of exercise in both males and females, whereas a reduction in the percentage of type IIB fibres was observed only in females, and female mice also exhibited a lower average fibre cross‐sectional area (CSA) with exercise.

Average fibre CSA (µm²), as well as CSA of type I, IIA, and IIB fibres, is presented for males (A) and females (C). Muscle fibre type composition is shown for males (B) and females (D). Red asterisks indicate a significant tumour main effect (C26 vs. PBS; P < 0.05), whereas blue asterisks indicate a significant PoWeR exercise main effect (PoWeR vs. SED; P < 0.05), as determined by two‐way ANOVA. *P < 0.05.

Figure A8.  Transcriptomic and GO enrichment analysis of skeletal muscle across sex, tumour burden, and PoWeR training.

(A) Overview of differentially expressed genes across groups (blue numbers: upregulated genes; red numbers: downregulated genes). (B–E) GO enrichment of biological processes and molecular functions in male and female mice comparing PBS, C26 sedentary, and PoWeR‐trained conditions, highlighting tumour‐induced alterations in metabolic and mitochondrial pathways and their modulation by exercise. Transcriptomic and GO enrichment analysis of skeletal muscle across sex, tumour burden, and PoWeR training. (A) Overview of differentially expressed genes across groups (blue numbers: upregulated genes; red numbers: downregulated genes). (B–E) GO enrichment of biological processes and molecular functions in male and female mice comparing PBS, C26 sedentary, and PoWeR‐trained conditions, highlighting tumour‐induced alterations in metabolic and mitochondrial pathways and their modulation by exercise. RNA sequencing analysis stratified by biological sex.

Data availability statement

Raw data from RNA Sequencing are available on GEO, under accession number GSE324100. All other raw data are available on FigShare: https://figshare.com/s/df170626f9761a44993c, https://figshare.com/s/3f72255ae2fa1d118bd3, https://figshare.com/s/cbecf80930de2362a3e3, https://figshare.com/s/becccb5238550aa926a6, https://figshare.com/s/b1dbe065a99fd5cbdd52.