Exacerbated central fatigue and reduced exercise capacity in early-stage breast cancer patients treated with chemotherapy

Fifteen patients with breast cancer were included in the study (patient group). Eligibility criteria included French-speaking, nonpregnant, ≥ 18 years old, women with Scarff‐Bloom‐Richardson grade I–III and early-stage breast cancer (1–3), World Health Organization (WHO) performance status between 0 and 2, within 2 weeks after completion of anthracycline-cyclophosphamide and taxane-based (neo)adjuvant chemotherapy.

Women were excluded if they had psychiatric, musculoskeletal, or neurological disorders.

Fifteen healthy volunteer women (control group) were matched to the characteristics of the breast cancer group regarding age, weight, height, and physical activity level. Eligibility and exclusion criteria were the same as those identified in the patient group except for criteria related to breast cancer status. Indeed, participants from the control group had no history of cancer or another chronic disease.

The characteristics of the participants included in the present study (NCT04639609) are presented in Table 1. All participants provided written informed consent before enrollment, and the study was conducted in accordance with both the Declaration of Helsinki and the ethics approval received from the national ethics committee (2020-A01272-37).

All participants carried out the experimental protocol, conducted as follows: (1) familiarization with isometric knee extensor contractions and the associated experimental procedures, followed 30 min later by (2) neuromuscular function assessment (i.e., preexercise), (3) the fatiguing exercise task, (4) neuromuscular function assessment throughout 10 min of recovery (i.e., postexercise) (Fig. 1).

Neuromuscular function of the knee extensors was investigated with femoral electrical nerve and transcranial magnetic stimulations. After a standardized warm-up including submaximal isometric knee extensor contractions, participants were asked to perform six 3 s-MVCs, separated by 30 s recovery to ensure full potentiation. Further attempts were made if the variability in MVC exceeded 5%. Stimulations were applied both during (superimposed twitch) and 1 s after (potentiated resting twitch, Q_tw) each MVC to assess voluntary quadriceps activation and muscle contractile properties, respectively (Nuzzo et al. 2018). For all attempts, strong verbal encouragement was given. Corticospinal excitability was measured during submaximal voluntary isometric contractions corresponding to 20% of each participant’s vastus lateralis maximal EMG root mean square (EMG-RMS) output obtained from MVC. We used submaximal contractions based on a constant EMG activity to counteract the confounding increase in corticospinal excitability associated with an increase in EMG during a fatiguing constant force exercise (Lévénez et al. 2008; Hoffman et al. 2009). These submaximal contractions corresponded to 26 ± 5% of the MVC force. Using visual feedback, participants were asked to perform three submaximal contractions for 5–10 s and separated by 30 s of recovery (McNeil et al. 2011). During each contraction, 3 transcranial magnetic stimulations were delivered.

Participants were then required to perform a fatiguing task consisting of 60 MVCs over a 5 min period (3 s contraction, 2 s relaxation) with strong verbal encouragement. A target line was set on the computer screen at 100% MVC to maximize performance, and an audio recording cued the start and stop of each MVC to follow the required duty cycle. Superimposed and potentiated resting electrical stimulations were elicited at first and every ten contractions to investigate neuromuscular fatigue development throughout the exercise. No magnetic stimulation was applied during the 60-MVC protocol.

After the fatiguing task, neuromuscular function was assessed immediately and at 1, 2, 3, 5, and 10 min to quantify recovery. MVC was performed with superimposed and potentiated resting twitches for each set, followed by a 20% EMG-RMS submaximal contraction with 3 superimposed transcranial magnetic stimulations.

The critical force during the 60-MVC protocol was the mean force recorded over the last six MVCs, while W’ was calculated as the total force impulse generated above the critical force (Burnley 2009). The total work done was also determined and calculated as the total force impulse generated during the 60-MVC protocol. Q_tw was calculated as the force amplitude between the baseline signal and the highest force value of the evoked potentiated twitch. Moreover, the following parameters of potentiated resting twitch, as indicators of muscle contractile properties, were also calculated: contraction time (CT), which is the time from the start of the contraction to Q_tw; half relaxation time (HRT), which is the time from Q_tw to 50% decline in Q_tw; maximal rate of force development (RFD), which is the steepest rate of torque development (i.e., highest positive derivative of the torque for an interval of 10 ms between two cursors placed on either side of the torque rise). Voluntary activation was calculated using the twitch interpolation method (Merton 1954), with the amplitude of the superimposed twitch (SIT) and Q_tw, using the following formula: Voluntary activation (%) = [1 − (SIT ÷ Q_tw) \(\times\) 100]. The peak-to-peak amplitudes of M_max and MEP were measured between maximum and minimum values for the electromyographic signals. To account for any alteration of muscle sarcolemma excitability and variability associated with EMG measurements (Lepers et al. 1985), the amplitude of each MEP was normalized to concomitant M_max (MEP.cM_max⁻¹), and RMS data were then normalized to the root mean square recorded during pre-exercise MVCs (RMS_%MVC)_. The 3 MEP.cM_max⁻¹ obtained during each submaximal contraction were averaged together to minimize the variability of this parameter.

The sample size calculation was based on a previous investigation documenting MVC in breast cancer patients at the end of the adjuvant treatment compared to healthy controls (Gomes et al. 2014). Assuming an effect size of 0.96, \(\alpha\) = 0.05, and \(\beta\) = 0.8, the minimum number of participants required to establish a significant difference in maximal voluntary force between groups was calculated at fifteen per group (G*power, version 3.1.9.4).

All statistical tests and graphs were generated with GraphPad Prism version 8 software (GraphPad Software, San Diego, California, USA). Data are presented as the means ± SD. The Shapiro–Wilk test and the Levene test were used to check for normality and variance homogeneity of the data, respectively. Paired-samples t-tests were used to compare participants’ characteristics (age, height, body mass, body mass index, fat-free mass, fat mass, and skeletal muscle mass), muscle architecture (muscle thickness, pennation angle, and fascicle length), questionnaires scores (i.e., GPAQ, FACT-G, and FACIT-F), mechanical parameters (i.e., critical force, W’, and total work done) and neuromuscular function parameters (i.e., MVC, RMS_%MVC, Q_tw, voluntary activation, M_max, MEP.cM_max⁻¹) between groups. Two-way ANOVAs with repeated measures (group \(\times\) time) were used to compare fatigue parameters. The time effect of the ANOVA included all data available of the parameter of interest pre, during and postexercise. Specifically, it included preexercise, MVC 1 to 60 and the recovery period (1, 2, 3, 5 and 10 min) for MVC and RMS_%MVC; MVC 1, 10, 20, 30, 40, 50 and 60 and the recovery period for Q_tw, voluntary activation, CT, HRT, RFD and M_max; preexercise values and the recovery period for MEP.cM_max⁻¹. Multiple comparison analysis was performed with the Sidak post hoc test when a significant difference was found. Statistical significance was set at P < 0.05.

Participants’ characteristics are reported in Table 1. Patients and healthy controls were matched for age, weigh, height, and physical activity level. Moreover, fat-free mass (P = 0.539), fat mass (P = 0.838) and skeletal muscle mass (P = 0.262) were not different between the two groups.

Preexercise MVC was lower in patients than in controls (− 15 ± 18%, P = 0.022, Fig. 2). Preexercise VA and Q_tw were not different between groups (P = 0.589 and P = 0.523, respectively, Table 1) as well as Q_tw associated mechanics (CT, P = 0.782; HRT, P = 0.179; and RFD, P = 0.637; Table 2). There was also no difference between groups on RMS_%MVC for the three muscle groups (VL: P = 0.853; VM: P = 0.407; RF: P = 0.504). M-wave amplitudes were lower in patients than controls (Table 2). VL and RF MEP.cM_max⁻¹ were higher, and VM MEP.cM_max⁻¹ was similar in patients compared to controls (Table 2).

As illustrated in Fig. 3, muscle thickness was lower in patients than controls (− 13.4 ± 14.8%, P = 0.004, Fig. 3A), while no difference was observed between groups for pennation angle (P = 0.515, Fig. 3B) and fascicle length (P = 0.362, Fig. 3C) in the vastus lateralis muscle.

Scores of cancer-related fatigue and quality of life are reported in Table 3. The FACIT-F score was significantly lower in patients than in controls (P < 0.001). Each item for quality of life was lower in patients than in controls (all P < 0.01), except for social well-being.

Preexercise MVC values did not differ from the first MVC of the exercise (patients: 303 ± 54 N vs. 297 ± 56 N, P = 0.732; controls: 368 ± 89 N vs. 365 ± 87 N, P = 0.919), indicating no pacing strategy by the participants at the start of the 60-MVC protocol.

Breast cancer patients showed a substantial reduction in exercise capacity at chemotherapy completion compared to their healthy counterparts. In this context, the 60-MVC exercise protocol was selected to provide a cohesive framework within which to investigate the mechanistic bases of this impairment in exercise capacity. Indeed, the reduced exercise capacity observed in early breast cancer patients, evidenced by a ~ 23% reduction in total work done, was characterized by lower maximal force and absolute critical force but similar W’. In other words, we observed a downward shift in the MVC curve in patients compared to controls (Fig. 5A). As a consequence, when expressed relative to maximal force (i.e., preexercise MVC), the critical force was similar in patients and controls, indicating that early breast cancer led to a proportional reduction in exercise capacity in the severe and the heavy intensity domains (Poole et al. 2016). However, we also documented a ~ 24% reduction in absolute critical force in patients compared to controls (Fig. 4A), which highlights the magnitude of the reduction of the greatest metabolic rate that results in wholly oxidative energy provision (Poole et al. 2016). In contrast, we found a similar magnitude of W’ between groups, a parameter that has been associated with the development of neuromuscular fatigue (Poole et al. 2016; Zarzissi et al. 2020) and the loss of muscular efficiency (Murgatroyd and Wylde 2011). This result is in contradiction with our hypothesis (i.e., greater W’ in patients than controls) but consistent with the observation of similar, instead of greater, level of neuromuscular fatigue between patients and their healthy counterparts. Therefore, these results provide novel insights from a bioenergetics perspective into the reduction in exercise capacity characterizing patients with early breast cancer.

The reduction in exercise capacity observed in breast cancer patients was associated with an exacerbated central fatigue, as evidenced by the reduction in voluntary activation measured immediately at exercise cessation, compared to their healthy counterparts (Fig. 6A). This result is supported by previous research suggesting, through deductive inferences (i.e., central fatigue was not directly measured), that the mechanisms related to exacerbated fatigue originated within the central nervous system in cancer survivors with solid tumors (Cai et al. 2014; Kisiel-Sajewicz et al. 2012, 2013; Yavuzsen et al. 2009). Mechanistically, greater central fatigue was likely not explained in the present study by reduced corticospinal excitability, as evidenced by the absence of difference between groups (Table 2). However, it is important to note that several studies documented that corticospinal excitability remained unaltered in the face of a decrease in voluntary activation following fatiguing exercise (Gandevia et al. 1996; Kalmar and Cafarelli 2006). Therefore, the lack of a net effect on corticospinal excitability does not necessarily designate the absence of change but a counterbalance of excitatory and inhibitory influences on the corticospinal pathway (Weavil and Amann 2018).

Importantly, greater central fatigue was observed in patients, while peripheral fatigue was similar compared to controls. Taken together, and given the tight relationship between intramuscular metabolic perturbation and peripheral fatigue (Hureau et al. 2022), these results might suggest overactive group III/IV muscle afferent feedback in patients with breast cancer, a phenomena that have already been evidenced in other chronic diseases (Gagnon et al. 2012; Amann et al. 2014). Indeed, metabo- and mechanosensitive group III/IV sensory neurons project from skeletal muscles to the central nervous system and promote central fatigue during intense exercise to restrict peripheral fatigue and protect the exercising muscles from severe threats to muscle homeostasis (Blain et al. 2016; Hureau et al. 2018). Further studies specifically investigating group III/IV muscle afferent feedback in patients with breast cancer are needed to evaluate whether this mechanism is involved in the central maladaptations observed in the present study.

The present study demonstrated that peripheral fatigue, as evidenced by the decrease in Q_tw, was similar in breast cancer patients compared to healthy controls (Fig. 6B). Based on this result, peripheral fatigue seems relatively preserved during chemotherapy. Interestingly, a recent study investigated the determinants of cancer-related fatigue by separating fatigued and nonfatigued cancer survivors into two groups based on a clinical cutoff point (Brownstein et al. 2022). This study documented that clinically fatigued patients experienced more peripheral fatigue during cycling exercise than those reporting no clinical fatigue. In contrast with the present study, this investigation demonstrates the existence of peripheral alterations in breast cancer survivors. This discrepancy between studies might be explained by the difference in the populations investigated (i.e., fatigued and nonfatigued patients vs. patients and healthy controls) and/or the exercise modality utilized to induce neuromuscular fatigue (i.e., whole-body cycling vs. single-leg extension).

Regardless, the observation of similar acute peripheral fatigue between patients and controls does not rule out the presence of chronic peripheral alterations subsequent to chemotherapy. Indeed, patients displayed a ~ 15% reduction in MVC at baseline compared to their healthy counterparts. Moreover, our results showed reduced skeletal muscle mass, as evidenced by lower vastus lateralis muscle thickness measured via ultrasonography, which is supported by a previous study showing a reduced muscle cross-sectional area in breast cancer patients compared to healthy controls via vastus lateralis muscle biopsies (Guigni et al. 2018b). In addition to this lower muscle quantity, impairments in muscle quality have been previously identified in breast cancer patients, such as a reduction in mitochondrial content (Mallard et al. 2022c). Therefore, profound chronic changes in both muscle quantity and quality are likely responsible for the reduction in MVC observed in patients compared to controls.

The present study investigated the etiology of neuromuscular fatigue in an homogeneous group of patients (i.e., from one cancer type) compared to healthy controls considering the large heterogeneity of symptoms, treatment options, and outcomes across cancers (Lin 2019; Hickok et al. 2005; Baracos et al. 2018). While the control group was matched one-to-one with a patient for age, weight, height, and physical activity level, patients displayed greater subcutaneous adipose tissue on the vastus lateralis muscle than controls, which might affect EMG recording (Farina et al. 2004). However, EMG data were normalized (RMS_%MVC) to allow relative comparisons between groups. In addition to MEPs, analyses of the silent period resulting from transcranial magnetic stimulations could have provided more information on intracortical inhibition. However, this study was not designed to measure the silent period as we did not give participants the specific instructions that are required to assess it rigorously (Hupfeld et al. 2020).