8- versus 12-week supervised resistance training with home-based physical activity during colorectal cancer treatment: a pilot randomized dose-comparison trial

Abstract

Background

Patients with colorectal cancer (CRC) undergoing systemic therapy frequently experience functional decline, fatigue, and treatment-related symptoms. Resistance training (RT) is safe during treatment, but the minimum effective duration remains unclear.

Objective

To compare the effects of an 8-week versus 12-week supervised RT program on chemotherapy-induced peripheral neuropathy (CIPN) and selected functional, psychosocial, behavioral, and hematological/inflammatory parameters in patients with CRC.

Methods

In this pilot randomized clinical trial, 30 adults with CRC were randomized (1:1) to 8 weeks (n = 15) or 12 weeks (n = 15) of supervised RT (2 sessions/week) plus home-based physical activity (3 days/week). The primary outcome was CIPN (EORTC QLQ-CIPN20). Secondary outcomes included handgrip strength, quality of life (EORTC QLQ-C30), anxiety/depression (HADS), physical activity, and hematological/inflammatory parameters.

Results

All participants completed the intervention, with adherence > 90% and no adverse events. No between-group differences were observed in CIPN. Handgrip strength suggested greater adjusted gains in the 12-week group for both dominant (pFDR = 0.026) and non-dominant hand (pFDR = 0.009). Psychological outcomes showed an adjusted between-group difference for HADS total score (pFDR = 0.022). Exploratory analyses suggested a reduction in NLR after 12 weeks (p = 0.043), with no significant adjusted between-group differences. Nominal between-group differences in fatigue and gastrointestinal symptoms were observed only in unadjusted analyses.

Conclusions

Supervised RT during systemic treatment in CRC appears feasible and safe. In this pilot dose-comparison trial, both programs were associated with improvements in selected outcomes, although efficacy cannot be established. A 12-week duration may be associated with additional benefits; however, findings are exploratory. Larger trials including usual-care controls are needed.

Registration

The clinical trial was registered in ClinicalTrials.gov (NCT07307690) on 1 January 2026.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12885-026-16245-2.

Introduction

Colorectal cancer (CRC) is among the most prevalent malignancies worldwide and remains a leading cause of cancer-related morbidity and mortality [1]. Although advances in early detection and oncological treatments have substantially improved survival, many patients experience a considerable burden of treatment-related adverse effects, particularly during active chemotherapy, which can severely compromise physical, psychological, and functional health [2, 3]. During chemotherapy, individuals with CRC commonly develop a multifactorial decline characterized by reduced muscle strength, cancer-related fatigue, emotional disturbances, decreased physical activity levels, and persistent gastrointestinal symptoms. These impairments frequently coexist, leading to diminished quality of life, reduced functional capacity, and increased vulnerability to physical deconditioning [4–6]. Notably, active chemotherapy represents a critical period during which functional decline may accelerate if not appropriately addressed [7].

In addition to these functional and psychological impairments, chemotherapy is often accompanied by a state of low-grade systemic inflammation and hematological alterations, which may contribute to symptom burden, fatigue, mood disturbances, and reduced physical capacity [8–11]. This inflammatory milieu has been increasingly recognized as a relevant pathophysiological mechanism underlying cancer- and treatment-related symptoms, and its modulation has emerged as a potential target for supportive interventions. In this context, exercise—particularly resistance training (RT)—has emerged as a safe and effective non-pharmacological strategy to mitigate many of the adverse effects associated with cancer and its treatments [12, 13]. RT plays a central role in oncology care due to its capacity to preserve muscle strength, attenuate functional decline, reduce fatigue, and improve psychological well-being, even during active treatment [14, 15]. Muscle strength, commonly assessed using handgrip strength, is a clinically meaningful and widely used functional marker and a strong predictor of health outcomes in patients receiving chemotherapy, with low values associated with increased treatment-related toxicity [16–18].

Beyond its functional benefits, exercise has been proposed as a potential modulator of systemic inflammation and immune function [19]. However, evidence regarding the effects of exercise on hematological and inflammatory markers during active chemotherapy remains limited and inconsistent [20], particularly in patients with CRC, and the influence of intervention duration on these physiological adaptations is largely unknown [21].

This issue is especially relevant in CRC, where fatigue, emotional distress, reduced physical activity, and gastrointestinal toxicity may persist or fluctuate across treatment cycles. From a physiological perspective, longer interventions may help counteract cumulative catabolic effects and consolidate neuromuscular adaptations [22], whereas from a behavioral standpoint, longer interventions may promote sustained engagement in physical activity [18]. Moreover, longer intervention periods may be necessary to induce detectable changes in systemic inflammatory markers, as immunological adaptations to exercise tend to occur more gradually than neuromuscular or functional adaptations [23, 24]. Conversely, prolonged interventions may increase patient burden and compromise adherence [25, 26], underscoring the need to identify the minimum effective duration that maximizes benefit while remaining feasible.

Most randomized controlled trials evaluating exercise in individuals with CRC have implemented programs ranging from 8 to 18 weeks in duration [27–33]. Available evidence indicates that relatively short interventions, typically around 6–8 weeks, are sufficient to elicit improvements in muscle strength and selected quality-of-life domains [27, 30]. However, it remains unclear whether extending RT beyond this initial period confers additional clinically meaningful benefits during active treatment, particularly with respect to symptom-related outcomes and underlying physiological responses, or whether the catabolic and symptomatic burden associated with chemotherapy limits further adaptations [22, 34].

Despite the growing interest in exercise oncology, to date, no randomized controlled trials in patients with CRC undergoing active chemotherapy have specifically compared different durations of the same RT program while holding other exercise variables constant. Although some studies in other cancer populations have reported outcomes at multiple time points during chemotherapy, they were not designed to specifically evaluate intervention duration [35, 36]. Consequently, evidence to optimize RT prescriptions during chemotherapy remains limited.

Therefore, the aim of the present randomized clinical trial was to compare the effects of an 8-week versus a 12-week RT program on muscle strength, emotional status, physical activity, chemotherapy-induced peripheral neuropathy, and quality of life in individuals with CRC undergoing chemotherapy, while exploratorily assessing changes in selected hematological and inflammatory parameters. It was hypothesized that both interventions would be beneficial, with the longer program potentially conferring additional advantages in selected outcomes. Importantly, this study was designed as a comparative evaluation of intervention duration (dose-comparison), rather than to determine the efficacy of exercise compared with usual care.

Materials and methods

Trial design and study setting

A randomized, parallel-group clinical trial was conducted to examine the effects of two supervised resistance exercise interventions of different durations in patients with CRC undergoing active oncological treatment. This trial was reported in accordance with the CONSORT 2025 Statement. The study followed a prospective design with two intervention arms and was carried out at the Salamanca University Healthcare Complex. All procedures were approved by the corresponding institutional ethics committee (nº 2025/03) and were conducted in accordance with the Declaration of Helsinki. Prior to participation, all individuals provided written informed consent. The clinical trial was registered in ClinicalTrials.gov (registration number NCT07307690, Registered on 1 January 2026).

Participants

Patients were recruited through the Oncology Department of the Salamanca University Healthcare Complex. Eligible participants were adults diagnosed with CRC who were receiving chemotherapy and/or immunotherapy at the time of inclusion.

Inclusion criteria were: age 18 years or older; confirmed diagnosis of CRC; ongoing systemic oncological treatment; no participation in structured exercise programmes during the previous eight weeks; and sufficient physical and cognitive capacity to safely perform exercise and complete the assessment procedures, as determined by the treating oncologist and research team based on clinical judgment.

Exclusion criteria included: medical contraindications to physical exercise (such as severe cardiovascular disease, limiting musculoskeletal conditions, or bone metastases); diagnosis of other active malignancies; and premature discontinuation of oncological treatment or withdrawal from the study due to intolerance or clinical deterioration.

Intervention

After baseline assessment, participants were randomly allocated to one of two supervised RT groups that differed exclusively in the duration of the intervention:

• G1: an 8-week supervised RT group

• G2: a 12-week supervised RT group

Both groups followed the same exercise protocol, delivered twice weekly under direct supervision by a qualified physiotherapist, together with a programme to promote physical activity at home 3 days a week. Each session consisted of a standardized warm-up, a RT phase targeting major muscle groups, and a cool-down period. Exercise selection, intensity, volume, and progression were individualized based on participants’ clinical status, perceived exertion, and tolerance, in accordance with current exercise-oncology recommendations [37]. Training intensity was set at approximately 70% of one-repetition maximum (1-RM), with progression implemented by increasing the load by ~ 10% when participants were able to complete the prescribed volume (three sets of 12 repetitions) in two consecutive sessions without excessive fatigue or adverse symptoms.

The structure and content of the sessions were identical in both groups, with the total duration of the intervention being the only variable manipulated. The supervised RT program consisted of two weekly sessions (~ 50 min each). Each session included six exercises targeting major muscle groups, performed in a circuit format. Participants completed three sets of 12 repetitions, with 30 s of rest between exercises and 90 s between sets. Exercise selection included multi-joint and single-joint movements such as squats, deadlifts, lunges, bench press, rowing, and shoulder exercises. The home-based physical activity program consisted of three weekly sessions (~ 20 min each), including two sets of five functional exercises (e.g., sit-to-stand, lunges, upper-limb resistance movements, and core exercises), performed with one-minute work and one-minute rest intervals. In addition, participants were encouraged to perform daily walking (~ 60 min), adapted to individual tolerance.

Adverse events were monitored throughout the intervention period during all supervised sessions. Any exercise-related symptoms or clinical incidents were recorded by the supervising physiotherapist.

Supplementary material 1 provides a comprehensive description of the intervention.

Variable and outcomes; type and measurement

Primary variable and outcome

The primary outcome of the study was chemotherapy-induced peripheral neuropathy (CIPN). Patient-reported neuropathic symptoms were assessed using the 20-item chemotherapy-induced peripheral neuropathy module of the European Organisation for Research and Treatment of Cancer Quality of Life questionnaire (EORTC QLQ-CIPN20) [38]. This instrument includes three domains addressing sensory, motor, and autonomic disturbances. Responses were recorded on a four-point Likert scale ranging from 1 (“not at all”) to 4 (“very much”), and raw scores were linearly transformed to a 0–100 scale, where higher values reflect increased severity of neuropathic symptoms. CIPN was defined as the primary outcome due to its high clinical relevance, prevalence, and functional impact; however, given the pilot nature of the study, all outcomes were interpreted within an exploratory framework.

Secondary variable and outcomes

Strength

Upper limb strength was evaluated through handgrip strength assessment using a hydraulic Jamar^® dynamometer (J00105, Lafayette Instrument Company, USA), widely recognized as a reference instrument for this purpose. Grip strength was measured bilaterally following standardized procedures based on the Southampton Protocol [39]. Participants were seated comfortably in a chair with a backrest, and the testing procedure was explained prior to assessment. For each hand, three maximal voluntary contractions were recorded, each lasting five seconds, with ten-second rest intervals between attempts. Verbal encouragement was provided throughout the test to ensure maximal effort. During measurement, participants maintained a standardized position: shoulder adducted and in neutral rotation, elbow flexed at 90 degrees, forearm in neutral alignment, and wrist positioned in slight extension (0°–30°). The second handle position of the dynamometer was used for all participants, except in cases of smaller hand size, where the first position was selected and documented. All assessments were conducted by the same evaluator to ensure consistency. The highest value obtained from each hand was used for statistical analysis.

Quality of life

Quality of life was assessed using the European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire (EORTC QLQ) [40]. The EORTC QLQ-C30, a cancer-specific instrument widely used in oncology research, was employed for this purpose. This questionnaire includes 30 items, of which 24 are organized into five functional domains (physical, role, emotional, cognitive, and social functioning), three symptom domains (fatigue, pain, and nausea/vomiting), and one global health status scale. The remaining items evaluate additional symptoms, including dyspnea, appetite loss, insomnia, constipation, and diarrhea, as well as perceived financial impact.

Scores for each scale and individual item were transformed to a standardized 0–100 scale. Higher scores on functional domains reflected better functional status, whereas higher scores on symptom domains indicated greater symptom burden. The questionnaire was administered prior to initiation of the intervention and subsequently at regular follow-up intervals throughout the study period.

Anxiety and depression

Anxiety and depressive symptoms were assessed using the Hospital Anxiety and Depression Scale (HADS), a questionnaire with demonstrated reliability and validity across clinical populations [41]. The instrument comprises 14 items, with seven items assessing anxiety (subscale A) and seven items assessing depression (subscale D). Each item was scored on a four-point scale ranging from 0 to 3, yielding separate scores for anxiety and depression.

Physical activity

Overall physical activity levels were assessed using the Spanish version of the International Physical Activity Questionnaire (IPAQ) [42]. This instrument consists of seven items that capture the frequency and duration of physical activities performed at different intensity levels during the previous week. Based on metabolic equivalent (MET) values, physical activity was categorized as light (< 600 MET·min/week), moderate (600–3000 MET·min/week), or vigorous (> 3000 MET·min/week). Light-intensity activity included activities such as walking for transportation, domestic tasks, or leisure purposes; moderate-intensity activity encompassed actions such as cycling, recreational sports, or carrying light loads; and vigorous-intensity activity involved high-demand exercises, including aerobic training, heavy lifting, or physically strenuous work. In addition, time spent in sedentary behavior during the preceding week was recorded to provide a comprehensive overview of participants’ activity patterns. Total physical activity was calculated as the sum of MET·minutes per week derived from light, moderate, and vigorous activity categories [43].

Hematological and inflammatory parameters

Blood samples were obtained as part of routine clinical care and analysed in the hospital’s central laboratory using standardized automated procedures. The following variables were extracted from the clinical laboratory report at each assessment time point: hemoglobin (Hb, g/dL), mean corpuscular volume (MCV/MCV, fL), C-reactive protein (CRP/CRP, mg/L), white blood cell count (WBC) (×10⁹/L), and the neutrophil-to-lymphocyte ratio (NLR), calculated as absolute neutrophil count divided by lymphocyte count (dimensionless). All determinations were performed in the hospital’s central laboratory using standardized automated procedures.

Randomization

Participants were randomly assigned to one of the two intervention groups using a computer-generated randomization sequence. Randomization was performed by an independent researcher not involved in participant assessment or intervention delivery to ensure allocation concealment.

Blinding

Due to the nature of the exercise intervention, blinding of participants and therapists was not feasible. However, outcome assessors and the statistician responsible for data analysis were blinded to group allocation in order to reduce assessment and analytical bias.

Data analysis

All statistical analyses were performed using R (version 4.4.1), employing the tidyverse, haven, rstatix, FactoMineR, and ggplot2 packages.

Continuous variables were summarized as mean ± standard deviation or median (interquartile range), depending on data distribution, while categorical variables were described using frequencies and percentages. Baseline comparability between groups was assessed using independent samples t-tests or Mann–Whitney U tests for continuous variables, and chi-square or Fisher’s exact tests for categorical variables.

Directionality of all patient-reported outcome measures was standardised prior to analysis to ensure consistent interpretation across outcomes. Symptom-related scales (including CIPN, HADS, and symptom domains of the EORTC QLQ-C30) were oriented such that higher scores consistently reflected greater symptom burden (worse outcomes). Functional scales (EORTC QLQ-C30 functional domains) were maintained in their original orientation, where higher scores indicate better functioning. Where necessary, scores were recoded to align interpretation across measures.

Primary analysis

Intervention effects were evaluated using analysis of covariance (ANCOVA), with post-intervention values as dependent variables and baseline values included as covariates. Models were additionally adjusted for clinically relevant covariates, including cancer stage, number of chemotherapy cycles, and months of disease evolution. ANCOVA was selected as the primary analytical approach due to its ability to improve statistical efficiency and adjust for baseline differences. Given the small sample size and two-timepoint design, ANCOVA was preferred over mixed-effects models due to its greater statistical stability and lower risk of model overparameterization in pilot studies.

Model assumptions (normality of residuals, homoscedasticity, linearity, and homogeneity of regression slopes) were assessed using graphical diagnostics and residual inspection, with no relevant violations observed. Effect estimates are reported as adjusted mean differences with 95% confidence intervals. Statistical significance was set at p < 0.05.

Multiplicity and inferential framework

To control type I error inflation, false discovery rate (FDR) correction (Benjamini–Hochberg) was applied to prespecified primary ANCOVA between-group comparisons. All other analyses were considered exploratory and were not adjusted for multiplicity. Given the number of outcomes and the small sample size, the study was not designed for confirmatory inference; therefore, non-adjusted results should be interpreted as exploratory and hypothesis-generating, with a risk of false-positive findings.

Alternative analytical approach

Change-score analyses (Δ = POST − PRE) were conducted as an alternative analytical approach to the primary ANCOVA models to evaluate robustness of findings under different statistical assumptions. These analyses were not intended as confirmatory inference.

Boxplots were used to visually represent the distribution of change scores (Δ) for each outcome by group.

Effect sizes

Effect sizes were calculated to complement p-values and assess clinical relevance. Cohen’s d was used for parametric comparisons, and rank-biserial correlation (r) for non-parametric comparisons. Effect sizes were interpreted as small (d ≤ 0.2 or r ≤ 0.1), moderate (0.2 < d < 0.8 or 0.1 < r < 0.5), or large (d ≥ 0.8 or r ≥ 0.5).

Categorical outcomes

IPAQ-derived physical activity categories were analysed using frequencies and percentages. Between-group differences were assessed using Fisher’s exact test, and within-group changes using McNemar’s test when appropriate. Agreement between time points was evaluated using Cohen’s kappa (κ). In cases of sparse data or quasi-complete separation, penalized logistic regression (Firth correction) was applied.

Power and pilot design

Given the pilot nature of the study, no formal sample size calculation was performed. The study was not powered to detect statistically significant between-group differences in primary or secondary outcomes. The main objectives were feasibility assessment and estimation of effect sizes for future definitive trials. Therefore, all estimates should be interpreted with caution, as limited sample size results in imprecise estimates, wide confidence intervals, and an increased risk of type II error. Importantly, null findings should not be interpreted as evidence of absence of effect, particularly for secondary outcomes where clinically meaningful effects cannot be excluded.

Confounding and interpretation

Although multivariable adjustment was applied in all primary models, residual confounding cannot be excluded due to baseline imbalances in disease severity, treatment exposure, and behavioural variables, which may have influenced observed effects despite statistical adjustment.

Results

A total of 40 participants were assessed for eligibility; 10 declined to participate, and none were excluded based on inclusion criteria. Thirty participants were randomized and equally allocated to the 8-week and 12-week intervention groups (n = 15 per group). All participants received the allocated intervention, with no dropouts, losses to follow-up, or treatment discontinuations. Consequently, all randomized participants were included in the final analysis. Participant flow is summarized in Fig. 1.

Fig. 1.

Flowchart

Feasibility outcomes were favourable. The recruitment rate was 75% (30/40), and retention was 100%, with all enrolled participants completing the intervention. Adherence to the supervised exercise sessions was high (> 90% in both groups). No exercise-related adverse events were reported during the intervention period.

Sample size

This study was designed as a pilot randomized controlled trial aimed at assessing feasibility and generating preliminary estimates of effect sizes to inform future trials. Accordingly, no formal sample size calculation was performed, and the study was not powered for confirmatory inference. Therefore, statistical significance should be interpreted in an exploratory context rather than as confirmatory evidence of effectiveness.

Sociodemographic information

The baseline sociodemographic and clinical characteristics of participants in the 8-week (8 W, n = 15) and 12-week (12 W, n = 15) groups are summarized in Table 1.

Table 1.

Baseline characteristics of participants in the 8-week and 12-week intervention groups

Variable	8 weeks group (n = 15)		12 weeks group (n = 15)
	No.	%	No.	%
Age (years) (p-value = 0.220)
Mean	59.6		64
SD	8.7		10.1
Gender (p-value = 0.272)
Male	10	66.7	6	40
Female	5	33.3	9	60
Weight (kg) (p-value = 0.910)
Mean	70.4		69.8
SD	12.8		16.5
BMI (kg/m2) (p-value = 0.790)
Mean	25.4		25.9
SD	4.4		4.3
Cancer type (p-value = NA)
Colon	15	100	15	100
Rectal	0	0	0	0
Cancer Stage (p-value = 0.258)
II	2	13.3	2	13.3
III	6	40	1	6.7
IV	7	46.7	12	80
Number of Cycles (p-value = 0.298)
Median	4		7
IQR	2.5–8		4.5–16
Type of treatment (p-value = 0.471)
CT	8	53.3	5	33.3
IT	1	6.7	3	20
CT + IT	6	40	7	46.7
Months of evolution (p-value = 0.406)
Median	6		18
IQR	5–33.5		10–31

SD standard deviation, IQR interquartile range, CT chemotherapy, IT immunotherapy, CT + IT chemotherapy plus immunotherapy, BMI body mass index

No significant differences were observed between groups for key continuous variables. Mean age was 59.6 ± 8.7 years in the 8 W group and 64 ± 10.2 years in the 12 W group (p = 0.220, t-test). Weight and BMI were comparable between groups (70.4 ± 12.8 kg vs. 69.8 ± 16.6 kg, p = 0.910; 25.4 ± 4.4 kg/m² vs. 25.9 ± 4.3 kg/m², p = 0.790). The number of chemotherapy cycles and months of disease evolution did not differ significantly (median 4 [IQR 2.5–8] vs. 7 [IQR 4.5–16], p = 0.298; 6 [IQR 5–33.5] vs. 18 [IQR 10–31], p = 0.406).

Categorical variables were also similar. The distribution of gender was comparable (66.7% male in 8 W vs. 40% male in 12 W, p = 0.272). All patients had CRC, so comparison by cancer type was not applicable. Cancer stage and type of treatment (chemotherapy (CT) or immunotherapy (IT), were comparable between groups (p = 0.258 and p = 0.471, respectively).

Overall, although no statistically significant differences were observed for several baseline variables, clinically relevant imbalances were present in disease stage, treatment exposure, and behavioural variables, which should be considered when interpreting the findings.

Changes in chemotherapy-induced peripheral neuropathy

CIPN was analysed as the predefined primary outcome, although interpretation was framed within the exploratory context of this pilot trial. CIPN outcomes were analysed to evaluate the effects of the intervention between the 8-week and 12-week groups. Analysis of covariance (ANCOVA) was used, with post-intervention scores as dependent variables and baseline values as covariates. Additional adjustments were made for cancer stage, number of chemotherapy cycles, and months of disease evolution.

The 8-week and 12-week groups were comparable at baseline, with no significant differences in clinical variables (all p > 0.05), indicating balanced groups before the intervention.

No statistically significant between-group differences were observed in any CIPN subscales after adjustment (Table 2).

Table 2.

Between-group differences in Chemotherapy-Induced Peripheral Neuropathy (CIPN) subscale scores estimated using analysis of covariance (ANCOVA). Post-intervention outcomes were adjusted for baseline values, cancer stage, number of chemotherapy cycles, and months of disease evolution. Estimates represent the adjusted mean difference between the 12-week and 8-week groups, with corresponding standard errors, 95% confidence intervals, and p-values

Variable	Estimate	SE	t	p-value	p-adj	CI low	CI high
Sensory	0.91	5.87	0.15	0.878	0.992	-11.39	13.21
Motor	0.009	2.67	0.003	0.997	0.997	-5.59	5.61
Autonomic	-1.95	2.93	-0.66	0.514	0.992	-8.09	4.19

For the Sensory scale, the adjusted group effect was not significant (estimate = 0.91, p = 0.878). For the Motor scale, no group effect was observed (estimate ≈ 0.01, p = 0.997). For the Autonomic scale, no significant differences were found (estimate = -1.95, p = 0.514).

Overall, ANCOVA results indicated that the intervention duration (8 vs. 12 weeks) did not significantly influence CIPN outcomes after controlling for baseline values and clinical covariates.

To complement the primary analysis, descriptive pre–post changes in CIPN scores were also explored in both groups (8 and 12 weeks). These analyses were conducted to describe within-group evolution over time and were not adjusted for baseline differences or covariates.

The 8-week and 12-week groups were comparable at baseline. There were no significant differences in the clinical variables (all p-values > 0.05), indicating balanced groups before the intervention.

Changes in Chemotherapy-Induced Peripheral Neuropathy (CIPN) scores were evaluated between baseline (PRE) and follow-up (POST) for both groups, 8-weeks and 12-weeks. No statistically significant differences were observed between groups in any of the CIPN scales (Table 3; Fig. 2).

Table 3.

Changes in CIPN 20 subscales (Autonomic, Motor, Sensory) between the 8-week and 12-week intervention groups

Variable	p-value	Difference (95%CI)	Effect size
Autonomic	0.341	0.00 (-11.11 to 0.00)**	0.18**
Motor	0.370	0 (-12.50 to 0.00)**	0.17**
Sensory	0.226	-10.11 (-26.00 to 6.4)*	-0.45*

* indicates mean (t-test, d-Cohen’s); ** indicates median (Wilcoxon test, rank biserial correlation)

Fig. 2.

Between-group comparison of chemotherapy-induced peripheral neuropathy (CIPN-20) subscale change scores (autonomic, motor, and sensory) in the 8-week and 12-week intervention groups. Boxplots show the distribution of POST–PRE changes, including median, interquartile range, and individual participant values. Positive values indicate worsening of symptoms, whereas negative values indicate improvement

For the Autonomic scale, median changes were 0 in both groups, with a median difference of 0.0 points (95% CI: -11.11 to 0.0; effect size = 0.179; p = 0.341). Although some individual patients showed percentage changes ranging from − 50% in the 8-weeks group to 33.3% in the 12-weeks group, the overall median change was 0 and the variability was high. Therefore, these extremes did not translate into statistically significant differences between groups.

For the Motor scale, median PRE-POST changes were 0 in both groups, with a median difference of 0 points (95% CI: -12.50 to 0; effect size = 0.167; p = 0.370). Percentage changes were highly variable and not interpretable due to many baseline values of 0, indicating heterogeneous responses among patients. Some showed improvement, but no consistent pattern was observed across the group.

For the Sensory scale, the mean PRE-POST change was − 15.0 points for the 8-weeks group and − 5.2 points for the 12-weeks group, resulting in a mean difference of -10.1 points (95% CI: -26.0 to 6.4; effect size = -0.452; p = 0.226). The mean percentage change was − 36.9% in the 8-weeks group and − 39.7% in the 12-weeks group, suggesting that although differences were not statistically significant, some patients in the 8-weeks group experienced meaningful improvement in sensory symptoms.

Overall, individual patient responses varied, particularly in the sensory and motor scales. Reporting the magnitude of change, confidence intervals, and percentage change provides a more complete picture of potential clinical benefit beyond p-values.

Changes in strength

We assessed changes in dominant and non-dominant hand grip strength before and after the intervention in both the 8-week and 12-week groups.

ANCOVA analysis showed that intervention duration was associated with differences in handgrip strength. The 12-week group suggested higher adjusted strength gains compared with the 8-week group in both dominant (β = 2.54 kg, p = 0.026) and non-dominant hands (β = 2.95 kg, p = 0.009), after adjustment for covariates (Table 4).

Table 4.

Between-group differences in hand grip strength between the 8-week and 12-week intervention groups estimated using analysis of covariance (ANCOVA). Post-intervention hand grip strength outcomes were adjusted for baseline strength (dominant or non-dominant hand, respectively), cancer stage, number of chemotherapy cycles, and months of disease evolution. Estimates represent the adjusted mean difference between the 12-week and 8-week groups, with corresponding standard errors (SE), t-statistics, 95% confidence intervals (CI), and p-values obtained from the ANCOVA models

Variable	Estimate (Group G2)	SE	t	p-value	p-adj	CI low	CI high
Dominant	2.54	1.01	2.52	0.021	0.026	0.43	4.66
Non-dominant	2.95	0.81	3.65	0.002	0.009	1.26	4.65

Within-group changes are presented for descriptive purposes (Fig. 3). Both groups showed improvements in grip strength, with larger gains generally observed in the 12-week group. However, these analyses were not adjusted for baseline differences or covariates and should be interpreted cautiously.

Fig. 3.

Between-group comparison of handgrip strength change scores (dominant and non-dominant hand) in the 8-week and 12-week intervention groups. Boxplots show the distribution of POST–PRE changes, including median, interquartile range, and individual participant values. Positive values indicate strength gains, whereas negative values indicate reductions in strength

Figure 3 shows the distribution of changes in grip strength (POST–PRE) for descriptive purposes.

Changes in health-related quality of life

We further assessed between-group differences in health-related quality of life using ANCOVA. No statistically significant between-group differences were observed across clinical, psychological, and health-related quality of life outcomes after adjustment for multiple comparisons (Table 5).

Table 5.

Between-group differences across clinical, psychological, and health-related quality of life outcomes estimated using analysis of covariance (ANCOVA). Post-intervention outcomes were adjusted for baseline scores, cancer stage, number of chemotherapy cycles, and months of disease evolution. Estimates represent the adjusted mean difference between the 12-week and 8-week groups (Group G2), with corresponding standard errors (SE), t-statistics, 95% confidence intervals (CI), and both raw and false discovery rate (FDR)-adjusted p-values. FDR correction was applied using the Benjamini–Hochberg procedure to account for multiple comparisons

Variable	Estimate	SE	t	p-value	p-adj	CI low	CI high
CIPN Sensory	0.91	5.88	0.15	0.879	0.992	-11.39	13.21
CIPN Motor	0.01	2.67	0.00	0.997	0.997	-5.59	5.61
CIPN Autonomic	-1.95	2.93	-0.66	0.514	0.992	-8.09	4.19
HADS Anxiety	-1.62	0.81	-1.99	0.061	0.385	-3.31	0.08
HADS Depression	-3.17	1.00	-3.16	0.005	0.097	-5.27	-1.07
Global health status	2.61	4.05	0.64	0.527	0.992	-5.87	11.08
Physical functioning	-0.69	1.97	-0.35	0.732	0.992	-4.82	3.45
Role functioning	-3.59	3.72	-0.97	0.347	0.992	-11.36	4.19
Emotional functioning	-0.33	4.30	-0.08	0.940	0.992	-9.32	8.66
Cognitive functioning	-7.41	8.06	-0.92	0.370	0.992	-24.29	9.47
Social functioning	1.85	6.08	0.31	0.764	0.992	-10.87	14.58
Fatigue	6.06	5.92	1.02	0.319	0.992	-6.33	18.45
Pain	-0.66	6.93	-0.10	0.925	0.992	-15.16	13.84
Dyspnoea	6.21	3.48	1.78	0.090	0.429	-1.07	13.50
Insomnia	6.35	10.07	0.63	0.536	0.992	-14.74	27.43
Appetite loss	2.26	4.55	0.50	0.625	0.992	-7.27	11.78
Constipation	0.78	9.08	0.09	0.933	0.992	-18.22	19.78
Diarrhoea	23.49	9.56	2.46	0.024	0.226	3.49	43.50
Financial difficulties	4.18	9.76	0.43	0.674	0.992	-16.25	24.61

Across all HRQoL domains, no significant adjusted between-group differences were observed after FDR correction. Although diarrhoea showed a nominally significant unadjusted p-value (p = 0.024), this effect did not remain significant after adjustment (p_FDR = 0.226).

Within-group changes in health-related quality of life were also examined to describe how each intervention group evolved over time (Table 6). These analyses were exploratory and aimed at characterising intra-group changes rather than testing between-group effects.

Table 6.

Changes in health-related quality of life between the 8-week and 12-week intervention groups

Variable	p-value	Difference (95%CI)	Effect size
Physical	0.070	6.67 (0.00 to 20.00)**	0.61**
Role	0.313	16.67 (0 to 16.67)**	0.53**
Cognitive	0.421	0.00 (0.00 to 16.67)**	0.51**
Emotional	0.462	8.33 (-16.67 to 16.67)**	0.50**
Social	0.656	0.00 (-16.67 to 16.67)**	0.40**
Financial difficulties	0.609	0.00 (0.00 to 0.00)**	0.40**
Global health status	0.347	0.00 (-16.67 to 8.33)*	0.35*
Fatigue	0.041	14.07 (0.62 to 27.52)*	-0.78*
Nausea	0.608	0.00 (-16.67 to 0.00)**	0.08**
Pain	0.318	0.00 (-16.67 to 33.33)**	0.53**
Dyspnoea	0.343	-0.00 (0.00 to 0.00)**	0.39**
Sleep disturbance	0.847	0.00 (0.00 to 33.33)**	0.46**
Appetite Loss	0.701	-0.00 (0.00 to 0.00)**	0.41**
Constipation	0.021	-33.33(-66.67 to 0.00)**	0.23**
Diarrhea	0.006	-33.33 (-33.33 to 0.00)**	0.22**

* indicates mean (t-test, d-Cohen’s); ** indicates median (Wilcoxon test, rank biserial correlation)

Data in bold are significant values

In the 8-week group, changes were observed in physical, role, and emotional functioning, as well as reductions in fatigue and constipation (all p < 0.05). In the 12-week group, improvements were observed in global health status, physical and social functioning, and reductions in diarrhoea (all p < 0.05).

However, these within-group findings should be interpreted with caution, as they do not account for baseline differences or between-group comparisons.

To complement the ANCOVA results, a difference-in-differences analysis was performed. Although some outcomes (fatigue, constipation, and diarrhoea) showed nominal statistical significance, these findings did not remain significant after adjustment for multiple comparisons and should therefore be interpreted with caution.

These effects are further illustrated in Fig. 4, which displays the distribution of individual pre–post changes in fatigue, constipation, and diarrhoea for both intervention groups.

Fig. 4.

Between-group comparison of change scores for fatigue, constipation, and diarrhoea in the 8-week and 12-week intervention groups. Boxplots show the distribution of POST–PRE changes, including median, interquartile range, and individual participant values. Positive values indicate worsening of symptoms, whereas negative values indicate improvement

Overall, both intervention durations showed similar patterns of change in health-related quality of life. The ANCOVA results indicate that these changes were not significantly different between groups, suggesting that intervention duration did not meaningfully influence HRQoL outcomes.

Changes in anxiety and depression

To evaluate the impact of the intervention on psychological well-being, changes in anxiety and depression were assessed using the Hospital Anxiety and Depression Scale (HADS), where lower scores indicate fewer symptoms.

Between-group differences in post-intervention HADS scores were analysed using analysis of covariance (ANCOVA), adjusting for baseline scores, cancer stage, number of chemotherapy cycles, and months of disease evolution.

No statistically significant between-group differences were observed for anxiety or depression after adjustment for multiple comparisons (Table 7). Although the HADS-depression subscale showed a nominal unadjusted difference (p = 0.005), this effect did not remain significant after FDR correction (p = 0.097).

Table 7.

Between-group differences in anxiety and depression (HADS) outcomes estimated using analysis of covariance (ANCOVA). Post-intervention scores were adjusted for baseline values, cancer stage, number of chemotherapy cycles, and months of disease evolution. Estimates represent the adjusted mean difference between the 12-week and 8-week groups, with corresponding standard errors (SE), t-statistics, 95% confidence intervals (CI), raw p-values, and false discovery rate-adjusted p-values (pFDR)

Variable	Estimate (G2)	SE	t	p-value	p-adj	CI low	CI high
HADS-Anxiety	-1.62	0.81	-1.99	0.061	0.385	-3.31	0.08
HADS-Depression	-3.17	1.00	-3.16	0.005	0.097	-5.27	-1.07
HADS-Total	-4.76	1.36	-3.49	0.002	0.022	-7.61	-1.91

Participants exhibited small and variable changes in HADS scores over the course of the intervention. Anxiety showed slight improvements in both groups, while changes in depressive symptoms were inconsistent, with no clear pattern favouring either intervention. Overall, total HADS scores showed minimal variation across groups.

These trends are illustrated in Fig. 5, highlighting that anxiety levels remained generally stable and depressive symptoms fluctuated slightly, with individual variability observed within each group.

Fig. 5.

Between-group comparison of HADS change scores (anxiety and depression) in the 8-week and 12-week intervention groups. Boxplots show the distribution of POST–PRE changes, including median, interquartile range, and individual participant values. Negative values indicate improvement in psychological symptoms, whereas positive values indicate worsening

Changes in physical activity

Changes in physical activity were assessed using IPAQ categories before and after the intervention.

At baseline, no significant differences were observed between groups in low and high physical activity levels. However, a significant difference was found in moderate activity (p = 0.035), with a higher proportion of participants in the 12-week (12 W) group classified as moderately active.

Following the intervention, notable within-group shifts were observed across IPAQ categories in both the 8-week (8 W) and 12 W groups (Table 8). In the LOW category, participants tended to transition toward higher activity levels in both groups. This change reached statistical significance in the 8 W group (p = 0.045) and approached significance in the 12 W group (p = 0.074), although agreement between baseline and post-intervention classifications was low (κ = 0 and κ = 0.194, respectively).

Table 8.

Distribution of participants across IPAQ activity levels before and after the intervention

Intervention				IPAQ LOW POST		Total
				NO	YES
8 W	IPAQ LOW PRE p-value = 0.045; k = 0	NO	Count	11 (73.3%)	0 (0%)	73.3%
		YES	Count	4 (26.7%)	0 (0%)	26.7%
12 W	IPAQ LOW PRE p-value = 0.074; k = 0.194	NO	Count	9 (60%)	0 (0%)	60%
		YES	Count	5 (33.3%)	1 (6.7%)	40%

Intervention				IPAQ HIGH MEDIUM		Total
				NO	YES
8 W	IPAQ MEDIUM PRE p-value = 0.228 ; k= -0.32	NO	Count	2 (13.3%)	3 (20%)	33.3%
		YES	Count	8 (53.3%)	2 (13.3%)	66.7%
12 W	IPAQ MEDIUM PRE p-value = 0.371; k = 0.348	NO	Count	4 (26.7%)	4 (26.7%)	53.4%
		YES	Count	1 (6.7%)	6 (40%)	46.7%

Intervention				IPAQ HIGH POST		Total
				NO	YES
8 W	IPAQ HIGH PRE p-value = 0.008; k = 0.069	NO	Count	5 (33.3%)	9 (60%)	93.3%
		YES	Count	0 (0%)	1 (6.7%)	6.7%
12 W	IPAQ HIGH PRE p-value = 0.683; k= -0.216	NO	Count	9 (60%)	4 (26.7%)	86.7%
		YES	Count	2 (13.3%)	0 (0%)	13.3%

For moderate physical activity, both groups showed partial transitions toward this category; however, changes were not statistically significant (8 W: p = 0.228, κ = −0.32; 12 W: p = 0.371, κ = 0.348), with low agreement between time points.

In contrast, a significant increase in high physical activity was observed in the 8 W group (p = 0.008), whereas no significant changes were detected in the 12 W group (p = 0.683). Agreement between baseline and post-intervention classifications was low in both groups, indicating substantial within-subject variability.

When comparing the proportion of participants who changed IPAQ categories between groups, no differences were found for low activity (p = 1.000). However, significantly greater reclassification was observed in the 8 W group for both moderate (p < 0.001) and high (p = 0.021) activity levels. These findings should be interpreted cautiously due to baseline differences, particularly in the MODERATE category.

Between-group differences in post-intervention physical activity levels were further evaluated using regression models adjusted for baseline values, cancer stage, number of chemotherapy cycles, and months of disease evolution (Table 9).

Table 9.

Multivariable-adjusted logistic regression analyses of post-intervention IPAQ physical activity categories by intervention group. Adjusted logistic regression models controlling for baseline values, cancer stage, number of chemotherapy cycles, and months of disease evolution. IPAQ LOW was analysed using penalized logistic regression (Firth method) due to quasi-complete separation. Confidence intervals for MODERATE and HIGH categories are not interpretable due to model instability and sparse data

Variable	Estimate (G2)	95% CI	p-value
IPAQ LOW (Firth)	0.23	−3.67 to 5.40	0.879
IPAQ MODERATE	0.72	— (inestimable due to instability)	0.563
IPAQ HIGH	−1.89	— (inestimable due to instability)	0.076

Due to sparse data and quasi-complete separation in the low activity category, a penalized logistic regression model (Firth method) was applied. The adjusted analysis showed no statistically significant between-group differences in the likelihood of being classified as low physical activity at post-intervention (β = 0.23, 95% CI − 3.67 to 5.40, p = 0.879).

For moderate and high physical activity categories, standard logistic regression models showed substantial instability, reflected in extremely wide confidence intervals due to sparse data and limited model convergence. Therefore, confidence intervals are not interpretable for these outcomes, and results are reported mainly in terms of effect direction and statistical significance.

No significant between-group differences were observed for moderate physical activity (p = 0.563). For high physical activity, a non-significant trend toward lower odds in the 12-week group was observed (p = 0.076).

Overall, adjusted analyses did not confirm statistically significant between-group differences in physical activity levels, suggesting that observed changes were primarily driven by within-group improvements rather than intervention duration.

Sedentary behaviour, assessed as daily sitting time (minutes/day), differed significantly between groups at baseline. Participants in the 12 W group reported higher sitting time compared with the 8 W group (median 480 vs. 240 min/day; p = 0.019).

Following the intervention, reductions in sitting time were observed in both groups, with a larger unadjusted between-group difference in change (36 vs. 153.3 min; p = 0.030). However, after adjustment using ANCOVA controlling for baseline values and clinical covariates, this between-group difference was no longer statistically significant (β = 49.6, 95% CI − 46.0 to 145.3; p = 0.291).

Changes in hematological and inflammatory parameters

To further explore the physiological effects of the intervention, within-group changes in hematological and inflammatory markers were analyzed over 8 and 12 weeks. Differences between post- and pre-intervention values were calculated for hemoglobin (Hb), mean corpuscular volume (MCV), C-reactive protein (CRP), leukocyte count, and NLR. Normality assumptions were assessed to select appropriate statistical tests, and effect sizes were estimated using Cohen’s d for parametric data or rank-biserial correlation (r) for non-parametric data.

In the 8-week group, changes in hematological and inflammatory parameters were small and not statistically significant. Hemoglobin, MCV, CRP, leukocyte count, and NLR showed minimal variation, indicating limited physiological response to the shorter intervention.

In contrast, the 12-week group showed slightly larger changes. Although most variables remained non-significant, modest increases were observed in Hb, MCV, CRP, and leukocytes. Notably, NLR decreased in the 12-week group in unadjusted analyses (p = 0.043); however, this finding should be interpreted cautiously given the exploratory nature of the analysis. These patterns are illustrated in Fig. 6.

Fig. 6.

Between-group comparison of change scores for hematological and inflammatory parameters (hemoglobin, mean corpuscular volume, C-reactive protein, leukocyte count, and NLR) in the 8-week and 12-week intervention groups. Boxplots show the distribution of POST–PRE changes, including median, interquartile range, and individual participant values. Positive values indicate increases over time, whereas negative values indicate decreases

Overall, within-group analyses indicated modest hematological changes, with a trend toward greater changes after 12 weeks was observed (Table 10).

Table 10.

Within-group changes in hematological and inflammatory parameters

	8 weeks			12 weeks
Variable	Difference (95%CI)	p-value	Effect size	Difference (95%CI)	p-value	Effect size
Hb	-0.05 (-0.6 to 0.5)*	0.857	-0.05*	0.43 (-0.35 to 1.20)*	0.255	0.31*
WBC	40.00 (-919.47 to 1427.47)**	0.887	0.04**	749.13 (-284.45 to 1782.72)*	0.142	0.4*
CRP	0.00 (-0.32 to 0.38)**	1.00	0.31**	0.18 (-0.60 to 0.95)**	0.099	0.43**
NLR	0.13 (-0.05 to 0.54)**	0.164	0.37**	-0.41 (-0.82 to -0.01)*	0.043	-0.57*
MCV	1.11 (-3.04 to 5.25)*	0.576	0.15*	3.07 (-0.39 to 6.52)*	0.078	0.49*

Hb Hemoglobin, CRP C-reactive protein, NLR Neutrophil-to-lymphocyte ratio, MCV Mean corpuscular volume, WBC White blood cell count

* indicates mean (t-test, d-Cohen’s); ** indicates median (Wilcoxon test, rank biserial correlation)

Data in bold are significant values

Adjusted between-group comparisons controlling for baseline values, cancer stage, number of chemotherapy cycles, and months of disease evolution are shown in Table 11.

Table 11.

Adjusted between-group differences in hematological and inflammatory parameters (ANCOVA with FDR correction)

Variable	Estimate	SE	t	p-value	p-adj	CI low	CI high
Hemoglobin	0.245	0.552	0.444	0.662	0.662	-0.911	1.400
C-reactive protein	0.402	0.456	0.882	0.389	0.486	-0.552	1.360
Leukocytes	960.0	935.0	1.030	0.317	0.486	-997.0	2917.0
Neutrophil-to-lymphocyte ratio (NLR)	-0.487	0.265	-1.840	0.082	0.410	-1.040	0.068
Mean corpuscular volume (MCV)	3.220	2.990	1.080	0.294	0.486	-3.020	9.470

No statistically significant between-group differences were observed for hemoglobin, CRP, leukocytes, or MCV after FDR correction. Although the NLR showed a trend toward lower values in the 12-week group, this did not remain statistically significant after multiple-testing adjustment (pFDR = 0.410).

Overall, adjusted analyses did not confirm significant between-group differences in hematological or inflammatory markers, suggesting that observed within-group changes were not primarily driven by intervention duration.

When comparing change scores between groups (Table 12), most variables showed no significant differences. NLR showed a nominal between-group difference in the change-score analysis (p = 0.010); however, this result was exploratory and should be interpreted with caution.

Table 12.

Changes in hematological and inflammatory parameters between the 8-week and 12-week intervention groups

Variable	p-value	Difference (95%CI)	Effect size
Hb	0.293	0.47 (-0.43 to 1.38)*	0.39*
WBC	0.152	495.13 (-999.42 to 1989.69)**	0.26**
CRP	0.245	0.14 (-0.69 to 0.97)**	0.21**
NLR	0.010	-0.66 (-1.14 to -0.18)**	0.47**
MCV	0.433	1.96 (-3.20 to 7.12)*	0.28*

Hb Hemoglobin, CRP C-reactive protein, NLR Neutrophil-to-lymphocyte ratio, MCV Mean corpuscular volume, WBC White blood cell count

* indicates mean (t-test, d-Cohen’s); ** indicates median (Wilcoxon test, rank biserial correlation)

Data in bold are significant values

Overall, modest hematological and inflammatory changes in both groups pointed to a potential anti-inflammatory effect with longer intervention. The reduction in NLR after 12 weeks, along with a nominal between-group difference in change scores, warrants an exploratory interpretation.

Adherence and adverse effects

Overall adherence to the interventions was excellent, exceeding 90% in all cases, and no adverse events related to physical exercise were reported in any participant.

Discussion

This pilot randomized dose-comparison trial examined whether extending the duration of a combined exercise intervention (8 vs. 12 weeks) was associated with differential outcomes in patients with CRC undergoing active chemotherapy. The primary objective was not to determine the efficacy of exercise per se, but to explore whether a longer exposure to a standardized intervention may be associated with additional benefits. Given the pilot design and small sample size, all findings should be interpreted as exploratory and hypothesis-generating, as the study was not powered for confirmatory inference. Importantly, the intervention was feasible, well tolerated, and associated with high adherence and no serious adverse events, supporting the safety of supervised RT during chemotherapy and aligning with previous literature [44–46].

No significant differences were observed between the 8-week and 12-week groups in CIPN after adjustment, although individual improvements were observed in some sensory dimensions. In the sensory subscale, a moderate effect size (− 0.452) was obtained, suggesting a trend toward improvement, with reductions of approximately 37% and 40% after the 8- and 12-week interventions, respectively. However, these findings should be interpreted cautiously given the variability and lack of statistical significance in adjusted analyses. Current scientific evidence regarding the effects of exercise on CIPN is heterogeneous. While some reviews indicate clinically relevant benefits of RT combined with balance and endurance [26, 43, 44], others suggest that the effects are modest and dependent on the intensity and duration of the programme [47, 48]. The high inter-individual variability observed in CIPN responses in the present study suggests heterogeneous sensitivity to exercise during chemotherapy. This variability may be influenced by factors such as baseline neuropathy severity, treatment exposure, disease stage, and individual physiological differences [49]. Given the pilot nature of the study and the limited sample size, these heterogeneous responses should be interpreted cautiously, and further research is needed to identify potential responders and non-responders to exercise interventions in this population [50].

Our findings suggest that increasing the duration of the training programme may not be sufficient to obtain clear benefits in CIPN, highlighting the need for trials specifically focused on this outcome. Baseline imbalances between groups, particularly in disease stage, treatment exposure, and sedentary behaviour, should be considered when interpreting the findings. Although statistical adjustment using ANCOVA was applied, residual confounding and regression to the mean cannot be excluded, especially given the small sample size. These factors may have influenced both the magnitude and direction of some observed effects.

Both groups showed improvements in muscle strength, with larger adjusted gains observed in the 12-week group. These findings suggest that longer training periods may be associated with greater strength gains. In the non-dominant hand, improvements were observed in both groups, with greater gains in the longer-duration intervention. Overall, these results indicate that an 8-week training period may be sufficient to increase muscle strength, whereas longer training durations may be associated with a greater magnitude of adaptation. These findings are consistent with previous studies showing that RT in cancer patients leads to improvements in muscle strength and physical function [13, 30, 51]. However, in the absence of a non-exercise control group, these findings should be interpreted as relative differences between intervention durations rather than evidence of efficacy.

One of the most relevant findings of this study was the observed changes in several quality-of-life parameters following the intervention. Nominal findings were observed in some symptom-related variables; however, these results did not remain significant after adjustment for multiple comparisons and should therefore be interpreted as exploratory. A trend towards improvement in the physical functioning of patients who participated in exercise programmes was observed, with no differences found between the two groups.

Fatigue showed a nominal between-group difference in the unadjusted change-score analysis, favouring the 8-week group; however, this finding was not confirmed in the adjusted ANCOVA analysis and did not remain significant after correction for multiple comparisons. Therefore, the fatigue results should be interpreted with caution and considered exploratory. Previous studies have reported beneficial effects of exercise on fatigue in patients undergoing cancer treatment [44, 52–54], but the present findings do not support a clear advantage of the 12-week programme for this outcome.

Nominal differences were observed in gastrointestinal symptoms, such as constipation and diarrhoea, suggesting that intervention duration may influence these outcomes. As this is a little-studied area, there is little previous literature on this population. Some research suggests that regular exercise can improve gastrointestinal function and motility in cancer patients [55, 56], which could explain the symptomatic benefits associated with a longer programme duration. In summary, a 12-week exercise programme may be associated with more favourable changes in some quality-of-life domains.

Psychological outcomes suggested a possible advantage of the 12-week programme, with a nominal between-group difference observed for depressive symptoms in unadjusted analyses and a moderate–high effect size (d = 0.82). In addition, an adjusted between-group difference was observed for HADS total score. However, these findings should be interpreted cautiously, particularly for depressive symptoms, as they did not remain significant after correction for multiple comparisons. These results may suggest that a longer duration of RT could be associated with improvements in mood-related outcomes during chemotherapy, although this hypothesis requires confirmation in adequately powered trials. Exercise may influence depression through multiple mechanisms, including reductions in fatigue, improvements in functionality, and enhanced overall well-being. These observations are consistent with previous literature supporting the benefits of physical exercise on depression in both general and cancer populations [57–61]. Given that depression in CRC is associated with lower treatment adherence and poorer quality of life, these preliminary findings warrant further investigation.

Anxiety did not show significant differences after the intervention, although a tendency toward improvement was observed. It is possible that other factors, such as disease progression or treatment-related uncertainty, may influence anxiety levels independently of the intervention. Previous studies have reported more consistent improvements following supervised exercise programmes [59–61], suggesting that additional factors may modulate this outcome.

An overall shift toward higher levels of self-reported physical activity was observed following the intervention, although these findings should be interpreted with caution due to baseline imbalances and the exploratory nature of the analyses. Some differences in transitions between IPAQ categories were observed between groups; however, these results were not confirmed in adjusted analyses and were characterized by substantial variability and model instability. Daily sitting time decreased in both groups; however, baseline imbalance and the absence of significant adjusted between-group differences limit the interpretation of this finding. Participation in supervised exercise programmes may act as a behavioural stimulus to promote more active lifestyles [62–64], although the present results do not support a clear duration-dependent effect.

The analysis of the hematological and inflammatory parameters suggested a reduction in NLR in the 12-week group in exploratory within-group analyses. However, adjusted between-group differences were not significant after correction for multiple comparisons. Therefore, these findings should be interpreted as hypothesis-generating. Although NLR has been proposed as an indirect index of inflammatory status, and previous studies have reported reductions following exercise interventions [65–68], the present results do not support a clear anti-inflammatory effect of longer-duration RT in this population.

Changes in other analytical parameters were modest. In general, greater improvements were observed in the 12-week group, with a tendency toward more favourable values, although these did not reach statistical significance. Hb, MCV, CRP, and leukocyte count showed trends toward improvement following longer training duration. These findings suggest that RT is safe from a haematological perspective during chemotherapy, consistent with previous studies [69]. It should be noted that the effects of exercise on analytical biomarkers are typically less pronounced than those observed in functional or quality-of-life outcomes. The inflammatory and catabolic burden associated with chemotherapy may attenuate the physiological adaptations induced by exercise, particularly in interventions of limited duration.

Taken together, the multidimensional changes observed across muscle strength, psychological well-being, physical activity patterns, and selected biological markers suggest that supervised RT during chemotherapy may influence several domains related to functional and physiological reserve. Although intrinsic capacity was not formally assessed, the observed changes are consistent with domains commonly included in this framework, such as locomotion, psychological status, and vitality. These findings support the potential relevance of exercise-based interventions to preserve multidimensional health during active cancer treatment, while highlighting the need for future studies specifically designed to evaluate intrinsic capacity in oncology populations. Research into the optimal duration of physical exercise interventions in cancer patients remains limited. Although longer programmes are often associated with greater improvements, there is no clear consensus regarding the minimum effective duration [70].

Implications of the study

The findings of this study support the feasibility and safety of RT during chemotherapy in patients with cancer. Both intervention durations were well tolerated and associated with high adherence, reinforcing the role of supervised exercise as a viable supportive strategy during active treatment. Although an 8-week programme appears sufficient to induce improvements in muscle strength, extending the intervention to 12 weeks may be associated with more consistent or greater gains in this outcome.

In contrast, findings related to gastrointestinal symptoms, fatigue, depressive symptoms, and inflammatory parameters should be interpreted as exploratory. Although some nominal differences were observed, these were not robust after adjustment for multiple comparisons and do not support definitive conclusions regarding the added benefit of a longer intervention duration. Similarly, no clear additional benefits were observed in CIPN with the extension of the programme, suggesting that other factors—such as type of exercise, intensity, or outcome measurement—may play a more decisive role in modulating this variable.

Future trials should aim to include larger and adequately powered samples, incorporate usual-care control groups, and explore the effects of combining different exercise modalities (e.g., resistance, balance, and aerobic training). In addition, further research should investigate underlying physiological mechanisms, including inflammatory and neuromuscular pathways, and evaluate the persistence of exercise-related adaptations beyond the intervention period.

Limitations

This study has several limitations that should be considered when interpreting the findings. First, the small sample size limits statistical power, resulting in imprecise estimates, wide confidence intervals, and an increased risk of type II error. Second, the absence of a usual-care control group precludes causal inference; therefore, observed changes may reflect natural adaptation to chemotherapy, regression to the mean, or non-specific effects rather than the intervention itself. Accordingly, results should be interpreted strictly as between-group comparisons (8 vs. 12 weeks), rather than as evidence of efficacy.

Third, baseline imbalances in disease stage, treatment exposure, sedentary behaviour, and physical activity may have introduced residual confounding, despite statistical adjustment. Fourth, the combined nature of the intervention (supervised RT plus home-based physical activity promotion) does not allow isolation of the specific effects of RT alone. Fifth, several outcomes were based on self-reported measures and may be subject to reporting bias.

Sixth, given the number of outcomes assessed and the limited sample size, there is an increased risk of type I error; therefore, findings should be interpreted as exploratory and hypothesis-generating. In addition, the absence of standardized functional status measures (e.g., ECOG performance status) may limit the characterization and generalizability of the sample. Body composition was not assessed, which may restrict the interpretation of potential morphological adaptations to the intervention. Finally, the absence of a follow-up period beyond the intervention precludes assessment of the long-term sustainability of the observed effects. Minimal clinically important differences were not evaluated, limiting the clinical interpretability of the findings.

Conclusion

Supervised RT appears feasible and safe in both 8-week and 12-week programmes in patients with CRC undergoing chemotherapy. A longer duration of the exercise programme was associated with greater improvements in muscle strength, whereas findings for gastrointestinal, fatigue, depressive, and inflammatory outcomes should be interpreted as exploratory. No clear differences were observed in CIPN. Exploratory within-group analyses suggested a reduction in NLR after 12 weeks, but adjusted between-group differences were not significant. These findings contribute to the understanding of the optimal ‘dose’ of exercise, although results should be interpreted as exploratory given the pilot nature of the study.

Abbreviations

Abbreviation

Definition

BMI

Body Mass Index

CIPN

Chemotherapy-Induced Peripheral Neuropathy

CONSORT

Consolidated Standards of Reporting Trials

CRC

Colorectal Cancer

CRP

C-Reactive Protein

CT

Chemotherapy

IT

Immunotherapy

G1

Group 1

G2

Group 2

Hb

Hemoglobin

IPAQ

International Physical Activity Questionnaire

IQR

Interquartile Range

MCV

Mean Corpuscular Volume

MET

Metabolic Equivalent of Task

NLR

Neutrophil-to-lymphocyte ratio

POST

Post-intervention

PRE

Pre-intervention

RT

Resistance Training

SD

Standard Deviation

WBC

White Blood Cell count

Authors’ contributions

C.M.S.: Conceptualization, Project administration, Writing—original draft and funding acquisition. L.P.F.: Writing—original draft, Investigation and Methodology. R.M.S.: Investigation, Methodology and Project administration. A.S.P.G.: Investigation, Methodology and Project administration. T.M.H.: Investigation, Methodology and Project administration. E.J.F.-R.: Investigation, Methodology and Project administration. N.A.G.: Formal analysis and Methodology. Y.L.M.: Investigation, Methodology and Project administration. S.L.E.M.: Investigation, Methodology and Project administration. E.F.S.: Investigation, Methodology and Project administration. L.H.M.: Investigation, Methodology and Project administration. M.F.: Investigation, Methodology and Project administration. J.L.S.-G.: Conceptualization, Investigation, Methodology, Project administration, Supervision, Writing—original draft and Writing—review and editing. All authors have read and agreed to the published version of the manuscript.

Funding

This research was funded by Colegio Profesional de Fisioterapeutas de Castilla y León, Spain (CPFCYL), S/2024/01503 CO. The funders were not involved in any aspect of the research, including its design, data collection and analysis, manuscript preparation, or the publication decision.

Data availability

The data-set can be obtained from the corresponding author on request.

Declarations

Ethics approval and consent to participate

The study was conducted in accordance with the Declaration of Helsinki and approved by the corresponding Institutional Research Ethics Committee of the Salamanca University Healthcare Complex (approval number: 2025/03). Informed consent was obtained from all subjects involved in the study.

Consent for publication

Not applicable. This manuscript does not contain any individual person’s data, images, or videos that could lead to the identification of participants.

Competing interests

The authors declare no competing interests.