Background
Central (respiratory) and peripheral (limb) muscle weakness is one of the main systemic effects of chronic obstructive pulmonary disease (COPD) [
1]. It primarily affects the lower limb muscles [
2], contributes to exercise intolerance [
3] and is associated with increasing disability and mortality [
4,
5].
Peripheral muscle weakness can be caused by both peripheral (muscle mass and contractile properties) and neural alterations [
6,
7]. Several studies have indicated that peripheral mechanisms do not account for all the strength loss in COPD. For example, lower quadriceps strength per unit of muscle cross-sectional area was found despite normal contractile properties in COPD patients [
8]. In addition, lower quadriceps strength is not observed in some patients when peripheral nerve stimulation is used instead of voluntary contraction [
9,
10]. This raises the hypothesis of altered neural drive to the muscle in COPD.
Surprisingly, the validation of this hypothesis remains quite controversial. By using the twitch interpolation method, most studies have failed to observe an activation deficit in COPD during maximal voluntary contractions [
11‐
13], while others have clearly shown this deficit during submaximal voluntary contractions [
14]. By using a different method and applying sensors directly to the scalp, lower neural drive from the motor cortex was reported during maximal and submaximal voluntary contractions in COPD [
15]. Both methodological concerns and the heterogeneity of muscle weakness in COPD could explain these discrepancies. First, the twitch interpolation method may be biased during maximal voluntary contractions, as the relationship between voluntary strength and the twitch-like increment in strength is no longer linear at high intensities [
16‐
18]. The measurement of voluntary activation by motor cortex stimulation could help to resolve this issue [
19]. Second, it is well known that muscle weakness is not ever-present in COPD patients since its prevalence is between 32 and 57% [
20,
21]. None of the aforementioned studies discriminated the patients with and without muscle weakness. Thus, different proportions of COPD patients with versus without muscle weakness, depending on the study, could have biased the overall results.
Furthermore, lower neural drive to the muscle can be ascribed to at least three mechanisms, which have never been questioned in the context of muscle weakness in COPD. First, decreased excitation from the brain is strongly expected in these patients given the decreased gray matter density in the motor and prefrontal cortex [
22,
23] and the presence of white matter lesions in the pyramidal neurons [
24]. Second, lengthened latency and lower amplitude of the maximal compound muscle action potential (Mmax) have also been described in COPD patients and suggest impaired neuromuscular transmission at the motor neuron and/or the motor plate level [
25‐
28]. Third, higher supraspinal inhibition during voluntary contraction could contribute to lowering the motor output by inhibiting the neural drive in COPD. Indeed, inflammatory mediators as well as lactate and protons [
29] induce increased activity in group III or IV muscle afferents, which acts supraspinally to limit motor cortical output [
30]. As COPD patients are known to exhibit high levels of chronic systemic inflammation [
31] and a predominance of type II glycolytic fibers leading to elevated muscle glycolytic activity [
32], the hypothesis of a motor output decrease due to increased inhibition is relevant.
The aim of the study was to address the question of a specific activation deficit in COPD with muscle weakness, and if so, by which mechanisms. We hypothesized lower motor cortex activation and excitability, higher corticospinal inhibition, and lower Mmax amplitude in patients with COPD and muscle weakness, as compared with COPD patients with normal muscle strength.
Discussion
The purpose of the study was to compare corticospinal and muscle functions in COPD patients with and without peripheral muscle weakness. Compared with patients with preserved muscle strength, patients with muscle weakness exhibited lower VAcortical during maximal voluntary contractions associated with a lower MEP/Mmax ratio. Conversely, no differences were found between patients with and without muscle weakness regarding silent period duration, central motor conduction time or M-wave properties.
Peripheral muscle alterations were assessed by QPt (non-voluntary contraction, reflecting both muscle structure and function) in this study. We found lower QPt in patients with muscle weakness compared with healthy controls and patients with normal muscle strength. These findings concur with those of a previous study highlighting the implication of peripheral alterations in the poor quadriceps strength of the weakest COPD patients [
21]. However, several clues are in agreement with a moderate role. First, when QPt was included as a covariate in ANCOVA analysis, the maximal voluntary contraction (QMVC) remained significantly lower in COPD
MW. Second, the percentage of QMVC variance explained by QPt was sharply lower in COPD patients with muscle weakness (62%) than in control subjects (80%). These data indicate that peripheral muscle alterations are less decisive in maximal force production in COPD patients with muscle weakness. Third, as shown in Fig.
4, the slope between QMVC and QPt was significantly lower in patients with muscle weakness compared with patients without muscle weakness and healthy controls. In other words, for a given QPt, the patients with muscle weakness tended to exhibit lower QMVC than the other groups. Collectively, these data point to the existence of an additional non-muscular limiting factor in COPD muscle weakness. The reduced neural excitability and the voluntary activation deficit that we found in parallel in the patients with muscle weakness are relevant explanations for the unexplained part of COPD muscle weakness.
The existence of a voluntary activation deficit has been a controversial issue in COPD. A first consideration is the relevance of the twitch interpolation method during maximal voluntary contractions to assess voluntary activation [
11‐
14]. In the current study, we provided an additional estimation of voluntary activation by directly stimulating at the motor cortex level. We observed lower VA
cortical in the patients with muscle weakness, despite no VA
peripheral differences. This is not the first time that a mismatch between VA
cortical and VA
peripheral has been reported over the knee extensor muscles [
42‐
44]. The absence or relatively poor VA
peripheral changes compared with the VA
cortical changes has been explained by a lack of sensitivity of the twitch interpolation method [
42,
44]. Indeed, this method cannot provide a direct indication of the amount of neural drive reaching the muscle because the assessment takes place at the muscle level [
45,
46]. Moreover, the twitch interpolation method presents a nonlinear relationship at high force levels, such that changes in the voluntary force elicit minimal changes in the superimposed twitch size [
16‐
18]. For example, in a previous study, a gain of 5.7% in VA
peripheral induced an average 20.4% increase in QMVC [
18]. Another consideration is that voluntary activation and neural activity have previously been assessed in COPD patients as a whole without discriminating the patients with and without muscle weakness [
11‐
15,
47]. At first glance, the absence of a significant difference in VA
cortical between the controls and the patients as a whole (Table
1) might have led us to conclude that an activation deficit was probably not involved in the reduced quadriceps strength in COPD. However, the patients with quadriceps muscle weakness exhibited a significantly lower VA
cortical compared with patients with preserved quadriceps strength. This result suggests that in many patients, reduced motor cortex activation is involved in quadriceps muscle weakness. A lower MEP/Mmax ratio was also noted in the patients with quadriceps muscle weakness. Changes in MEP/Mmax can reflect changes in spinal or cortical excitability [
48]. In the current study, spinal excitability (Hmax/Mmax) did not differ between groups. Thus, the lower MEP/Mmax ratio suggests reduced motor cortex excitability in the patients with quadriceps muscle weakness, which supports the hypothesis of reduced voluntary activation from the motor cortex in these patients. Importantly, the only difference in pulmonary and blood gas data between the two groups of patients concerned the resting PaO
2 levels, which were significantly lower in the weak patients. To check that the VA
cortical alteration was not biased by hypoxemia differences, we performed an ANCOVA with PaO
2 as covariate. VA
cortical remained significantly lower in the patients with muscle weakness (F = 5.87,
p < 0.05). Furthermore, any oxygen desaturation during the quadriceps contraction of the study protocol was unlikely, since in a previous study where SpO2 was measured during similar efforts, the mean variation of SpO
2 was only 0.01 and non-significant [
15]. Consequently, the significantly lower VA
cortical in the weak patients can be considered as a consistent result.
In this study, we also aimed to assess the mechanisms potentially involved in the decreased voluntary activation in COPD. Silent period duration is thought to reflect the level of corticospinal inhibition [
49]. In accordance with a previous study, we found a lengthened silent period in COPD patients compared with healthy controls [
50]. However, the increased silent period duration in both groups of COPD patients regardless of muscle weakness necessarily indicates that higher corticospinal inhibition is not responsible for most of the observed loss of voluntary strength in the patients with muscle weakness.
Peripheral neuropathy has been widely described in COPD and has mainly been characterized by lower peripheral nerve conduction velocities and lower Mmax amplitude [
25‐
28]. In the current study, we found no differences regarding Mmax amplitude and latency between the two groups of patients and the healthy controls. These results suggest the preservation of neuromuscular transmission at the motor neuron and/or the motor plate level in COPD. This is also supported by the comparable central motor conduction time (corresponding to the MEP latency) between groups, which depends on both motor neuron and corticospinal conduction velocities. Conversely, we noted lengthened spinal reflex latency (Hmax latency) in the patients with COPD compared with healthy controls, regardless of muscle weakness. The alteration in Hmax latency without any central motor conduction time changes could be explained by impairment at the Ia afferent pathways, as the other pathways traveled by the H-reflex are the same as those traveled by the MEP. Therefore, these results suggest a selective alteration in the quadriceps Ia afferent pathways in the patients with COPD, which is in agreement with previous studies reporting greater impairment in the fascicles of sensory nerves than in motor nerves in these patients [
25]. Moreover, the lengthened Hmax latency regardless of muscle weakness also indicates that the alterations in the sensory nerve fascicles are not responsible for most of the observed loss of voluntary strength in the patients with muscle weakness.
In sum, higher corticospinal inhibition and impaired neuromuscular transmission are unlikely to be involved in the reduced quadriceps strength of the patients with muscle weakness. The most likely explanatory mechanism of the neural component of peripheral muscle weakness, and thus for reduced voluntary activation, is decreased excitation from the brain, which is supported by the observation of lower gray matter density in the motor cortex (precentral gyrus) in COPD [
22].
Although not a direct objective of the current study, our results may provide some new insights in the potential trigger for brain impairment in COPD. Several etiological factors have been advanced to explain brain impairment in COPD, the most important of which are cerebral vascular disease, inflammation, oxidative stress, smoking, hypoxemia and non-rapid eye movement (NREM) sleep desaturation [
51,
52]. First, we did not find any differences in smoking history between the two groups of COPD patients. Thus, a major implication of cigarette smoke in the reported brain alterations is unlikely. Furthermore, without being able to definitely discard the implication of hypoxemia, our results show that the reduced cortical activation was independent of PaO
2 levels.
Beyond mechanistic perspectives, the results of the study open new horizons for muscle weakness management in patients with COPD. Indeed, some specific interventions are known to promote neural adaptations such as eccentric exercise [
53], neuromodulation [
54] or electrical stimulation strength training [
55]. These interventions might help to improve responses to rehabilitation in COPD patients with muscle weakness from cortical origin. This is particularly relevant for efforts to improve adaptations to pulmonary rehabilitation programs, given the rate of patients who respond poorly to the classical programs [
56].
Methodological considerations.
We used a figure-of-eight coil to stimulate the leg motor cortex. Although a double-cone coil is more conventional to stimulate the leg motor cortex as the magnetic stimulus penetrates less deeply for a given intensity [
57], the figure-of-eight coil takes the advantage to be more focal [
58], and there is no clear rationale to give priority to one over the other in the literature [
57]. Apart from these considerations, some strict procedures helped us to increase the validity of the data provided by the figure-of-eight coil. First, to avoid potentially suboptimal output, we used a higher intensity (mean stimulator output was 96%) than the intensities usually reported with a double-cone coil (≈60–70%). We also excluded from the TMS data analyses the participants with no MEP plateau before reaching the maximal stimulator output (26.6%). This rate of exclusion was not much higher than in previous studies. For example, in younger healthy subjects, a rate of exclusion between 15 and 20% was reported [
59,
60]. In addition, the TMS voluntary activation data were quite consistent with the literature data. In the current study, the mean TMS voluntary activation using a figure-of-eight coil was 89% in the healthy elderly controls and 92% in the patients without muscle weakness. In a previous study using a double-cone coil on the same muscle group and on a similar population (healthy elderly), mean TMS voluntary activation was 90% [
61].
Another limitation was that the antagonist biceps femoris Mmax could not be measured in the study. The minimization of antagonist muscle activity during TMS is important to provide appropriate TMS data [
41]. However, the stimulation of the sciatic nerve to elicit biceps femoris Mmax is very challenging for technical and ethical reasons. First, the site of stimulation is surrounded by large muscles, which makes it difficult to avoid stimulating a muscle directly. Furthermore, to evoke Mmax on the biceps femoris, a very high stimulation intensity on the sciatic nerve (over 600 mA) is needed [
19]. This intensity is more than six times higher than that used for stimulating the femoral nerve in the current study. The use of such painful intensity is inconceivable in patients with chronic illness. Nevertheless, several strategies were employed to minimize antagonist activity during TMS. First, the skin impedance was rather null in both groups (< 5 kΩ). Second, we systematically checked to obtain the smallest possible response on the antagonist muscle by moving the coil by small amounts. Last, the study included a control group and, given the two aforementioned procedures, there is no reason to believe that the antagonist was more activated in one group than the others.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.