Defective Motor Control of Coughing in Parkinson’s Disease

SubjectsTwenty-three patients with idiopathic Parkinson’s disease (16 males, 7 females) age 55 to 79 yr (mean 64) characterized by different types of clinical expression of the disease and by different levels of clinical disability (16) were recruited (Table 1). Fifteen patients denied a previous smoking history; four patients (3 males, 1 female) had not smoked over the last 10 yr. None of them had a previous history of respiratory disease, and all were on levodopa-carbidopa substitution therapy at the time of the study. Twenty-three healthy, age-matched, nonsmoker volunteers (17 males, 6 females) age 47 to 78 yr (mean 65.7) who coughed in response to preliminary fog inhalation challenges served as a control group. None of the participants had suffered from respiratory tract infections in the preceding 6 wk. The experimental protocol adhered to the Recommendations of the Declaration of Helsinki for Human Experimentation. Individual informed consent was obtained after detailed explanations of the procedures, but not of the purposes, of the study.

Routine pulmonary function tests were performed in both patients and control subjects; they included spirometry and functional residual capacity measurements (gas dilution method). Reference values were taken from Morris and Koski (17). Respiratory function data of patients and control subjects are summarized in Table 2.

Experimental Procedures and ProtocolExperiments were carried out according to procedures previously described (18). Reflex cough (RC) was induced by inhalation of fog produced by a MIST-O₂-GEN EN143A ultrasonic nebulizer (MIST-O₂-GEN Equipment Company, Oakland, CA) whose output could be progressively increased in steps corresponding to 5% of the maximal attainable output level. The range of nebulizer outputs employed in the present study was from 0.08 to 4.45 ml/min (18). Maximal voluntary cough (MVC) efforts were obtained by repeatedly encouraging each participant to cough as forcefully as possible. The force of expiratory muscles was measured, by using standard equipment and techniques, as the maximal expiratory pressure (Pe _max), i.e., the highest pressure generated by a subject against a closed airway and sustained for at least 1 s after a full inhalation to near total lung capacity. During these measurements the cheeks and the floor of the mouth were supported with the palm of the hands by an investigator; participants were also vigorously urged to exhale as hard as possible for the entire maneuver which lasted at least 4 s. During all these experimental maneuvers, the electromyographic (EMG) activity was recorded from the abdominal muscles using surface Ag-AgCl electrodes positioned 3 cm apart along the line of right obliquus externus fibers, with the lower medial electrode 10 to 20 mm lateral to the edge of the rectus sheath and just above the level of the umbilicus. The EMG activity recorded with these electrodes during cough was considered to reflect the activation of the obliquus externus muscle, as well as the activity of deeper abdominal muscles with minimal contamination of the EMG signal by the rectus abdominis electrical activity (19). The EMG signals were differentially amplified (×2,000), bandpass filtered (50 to 1,000 Hz), full wave rectified, and passed through a “leaky” integrator (low-pass resistance-capacitance filter, time constant 50 ms) to obtain the so-called “integrated” EMG activity (IEMG). The IEMG activity was fed to a DC amplifier whose gain could be adjusted to obtain, for each subject, paper recordings of such an amplitude to allow accurate measurements. Before each challenge, subjects were asked to change their posture (trunk flexion) and simulate events such as exhalation of a long, audible breath and throat-clearing; the IEMG wave forms evoked by these maneuvers were compared with those recorded during voluntary coughing for differentiation.

Participants were comfortably seated on a dentist’s chair provided with head and arm rests and were repeatedly reminded to relax and breathe normally with as constant a pattern as possible. To facilitate electrodes positioning and to prevent the development of EMG activity of postural origin in the abdominal muscles (20), the back of the chair was tilted backward by approximately 30 degrees. In these conditions, abdominal muscles displayed no obvious rhythmic or tonic activity. Then, patients were requested to cough as forcefully as possible to record IEMG activity from abdominal muscles during MVC. Given the high variability of IEMG amplitudes observed during MVC, each participant was asked to perform 15 to 20 MVC maneuvers separated by a 5 to 10-s interval. No attempt was made to control the lung volume at which maximal expiratory thrusts were started. Three to five Pe _max maneuvers were also performed. No abdominal muscle activation was observed during the inspiratory phase that preceded expiratory efforts, both during MVC and Pe _max. Following a 10-min recovery period, each subject was connected to the nebulizer via a mouthpiece and inhaled during resting tidal breathing increasing fog concentrations obtained by adjusting the nebulizer output. To avoid rebreathing, an outlet proximal to the patient port of the apparatus was provided. In all instances, the fog inhalation time was standardized at 1 min for each nebulizer output; 2 to 3 min of rest were scheduled between steps. On appearance of cough, the test was discontinued and subjects were allowed to recover for at least 30 min to avoid tachyphylaxis. Then, the challenge was restarted with inhalation of the fog produced at the output step immediately below the first cough threshold. If cough could be elicited again at the same level that had previously been shown to evoke cough, the challenge was discontinued and that level taken as the subject’s cough threshold. Conversely, if no cough response could be obtained, the challenge was resumed and continued until cough could be elicited twice at the same nebulizer output. In only four experiments (three patients, one control subject) we found a difference between cough threshold observed on the first and second occasion. Thus, cough threshold was taken as the lowest fog output capable of evoking at least one cough during two distinct challenges separated by a time interval of approximately 30 min. This procedure ensured that the cough recorded was a reflex response to the challenge rather than a random event (18). During the assessment of cough threshold, the number of expiratory thrusts varied from to 2 to 6 in each participant.

Recordings of the studied variables were performed by means of an eight-channel chart recorder (HP 7758A, Hewlett and Packard, Palo Alto, CA; paper speed from 5 to 25 mm/s); the signals of the variables recorded during coughing were also fed to a FM tape recorder (HP 3960) for later analysis.

Data Collection and AnalysisAs fully described elsewhere (18), we measured on paper recordings at relatively high speed (25 mm/s) the peak of IEMG activity (IEMG_P) in arbitrary units (AU) during Pe _max maneuvers, MVC, and during each of the two fog inhalation periods required to assess cough threshold.

Measurements of IEMG amplitudes recorded in different experimental sessions cannot reliably be used for within- and between-subject comparisons without adequate processing because they are affected by several factors. These factors, which are related to subjects’ and recording conditions, include muscle size (which may vary considerably with the gender and age of the subject), the efficiency of skin-electrode coupling, skin resistance, electrode position, distance between electrodes, and adjustment of signal amplification. Thus, when expressed in AU, IEMG_P is a valid measurement only within a given subject and within the same experimental session; when considering different subjects or sessions, similar IEMG_P may actually correspond to markedly different levels of muscle activation and, hence, muscle forces. The conventional method to overcome this problem consists of normalization of IEMG amplitudes by assigning an arbitrary value, usually 1 or 100, to the highest achievable amplitude (absolute maximum), and to scale all other IEMG amplitudes accordingly, i.e., expressing them as a fraction or percentage of the absolute maximum. In our experiments, the highest IEMG_P was consistently attained during MVC. Thus, all IEMG_P values recorded in each participant during Pe _max maneuvers, MVC, and RC should have been expressed in relative units (RU), i.e., as a fraction of the highest IEMG_P recorded during MVC which can be confidently considered as the absolute maximum in our experimental conditions. This normalization would be adequate to compare IEMG_P values recorded in normal subjects, but it cannot be used to compare IEMG_P values between normal subjects and Parkinson’s disease patients.

In fact, due to the well-established relationship between IEMG activity and muscle force (18, 21), subjects who develop muscle weakness due to a central deficit in muscle activation and, therefore, in motor unit recruitment, should also display a correspondingly reduced IEMG activity. Indeed, expiratory muscle force is reduced in Parkinson’s disease patients (10), as confirmed by their lower Pe _max (see Results); if this reduction in expiratory muscle force is disregarded and IEMG data normalized as described previously, then the normalized IEMG_P values observed in patients and control subjects during Pe _max, MVC, and RC would not significantly differ. This is in sharp contrast with the finding that expiratory muscle force is reduced in our patients.

Therefore, due to the previously mentioned relationship between force and IEMG activity, the amplitude of the abdominal IEMG activity also should be accordingly reduced. In other words, while the maximal recorded IEMG_P activity can be confidently assumed to closely approach their absolute maximum in normal subjects and used for normalization, the maximal IEMG_P recorded in patients with a central deficit in motor activation is not their absolute maximum and cannot be used for normalization. In fact, the maximal IEMG_P recorded in Parkinson’s disease patients is the maximal IEMG_P they can attain in their present clinical condition, and it represents only a fraction of their absolute maximum. Therefore, to appropriately compare IEMG_P values in patients and control subjects, we need to obtain a reliable estimate of the absolute maximum IEMG_P in the patients. Given the well-known relationship between IEMG activity and muscle force, we can estimate patients’ absolute maximum IEMG_P by considering the extent of the reduction in their expiratory muscle force. At variance with other muscle groups, the maximal force that can be developed by the expiratory muscles has been determined in a wide number of subjects and the corresponding predicted values, which take into account differences in age and gender, are also available (22). Thus, the ratio of measured-to-predicted Pe _max can be taken as a reliable quantification of changes in expiratory muscle force and, hence, in the corresponding IEMG_P values. In fact, assuming that in each subject the proportionality between IEMG activity and force is maintained, it follows that:

All but one of the 23 patients tested coughed in response to fog inhalation. Cough threshold values ranged from 0.4 to 3.36 (median 0.87) ml/min in control subjects, and from 0.4 to 4.0 (median 1.31) ml/min in patients (Figure 1). The corresponding first and third quartiles were 0.4 and 1.31 ml/min in patients and 0.77 and 1.77 ml/min in control subjects, with interquartile ranges of 0.91 and 1.04 ml/min, respectively. Comparison between cough threshold values in the two study groups failed to show any significant difference.

Patients’ Pe _max, expressed as a percentage of predicted values, was 66.88 ± 11.95 (Figure 2); this value was significantly lower (p < 0.01) than in control subjects (98.01 ± 8.10). As shown in Figure 3A, there was a negative correlation (r_s = −0.88, p < 0.01) between Pe _max and the level of clinical disability.

In control subjects, IEMG_P values during MVC slightly but significantly (p < 0.05) exceeded those recorded during Pe _max, and RC (Table 3); in contrast, patients’ IEMG_P values observed during Pe _max maneuvers, MVC, and RC did not significantly differ, but were significantly lower than the corresponding values recorded in control subjects (p always < 0.01). In patients, the duration of Tec during RC turned out to be significantly longer (p < 0.01) not only than that observed during MVC, but also than that recorded in control subjects during both RC and MVC. On the other hand, Tec values recorded in control subjects during MVC and RC were similar. As a consequence, both in patients and controls, the rate of rise of IEMG activity (IEMG_P/Tec) during RC was significantly lower (p < 0.01) than during MVC; moreover, IEMG_P/Tec values during MVC and RC were significantly lower in patients than in control subjects (p always < 0.01; Table 3). However, when, compared with MVC, the rate of rise of IEMG during RC was significantly less steep in patients than in control subjects (p < 0.01). Furthermore, patients’ IEMG_P and IEMG_P/Tec during RC were inversely related (r_s = −0.86 and −0.85, respectively, p always < 0.01) to the level of clinical disability (Figures 3B and C).

Average patterns of ramp IEMG activity of the abdominal muscles during MVC and RC, both in patients and control subjects, are diagrammatically illustrated in Figure 4. As shown by the original recordings reported in Figure 4, as a rule, not only the rate of rise of IEMG activity, but also its decay was slower during MVC and RC in the patients. The time course of the decay of IEMG activity was not quantitatively analyzed.

The most important finding of our study is that the intensity of voluntary and reflex cough efforts, as indexed by the peak and rate of rise of abdominal IEMG activity, was significantly lower in patients with Parkinson’s disease than in age-matched control subjects, whereas cough threshold, although slightly higher in patients, was not statistically different in the two groups (Figures 1 and 4, see also Table 3). Another interesting result is that Pe _max, as well as IEMG_P and IEMG_P/Tec during RC, were inversely related to the level of clinical disability in Parkinsons disease patients (Figure 3).

These findings suggest that coughing is impaired in patients with Parkinson’s disease, and also indicate that, at least in patients in the less advanced stages of the disease, motor rather than sensory components of the cough reflex are primarily involved. Furthermore, they confirm that the force generated by the expiratory muscles is reduced in these patients (10), as revealed by data on Pe _max (Figure 2). The reduced Pe _max in Parkinson’s disease patients has been shown to reflect expiratory muscle weakness (10). In keeping with this conclusion, studies on the flexor and extensor muscles of the superior limb in these patients have demonstrated a reduction in peak muscle force and in the rate at which it was achieved (24). The main reason for muscle weakness appears to be a decrease in EMG activity of the agonist muscles. Peripheral changes in muscles or nerves are unlikely to explain muscle weakness because they cannot account for the finding that the strength of both respiratory and limb muscles is increased by antiparkinsonian treatment (10, 24). Thus, it seems plausible that expiratory muscle weakness in Parkinson’s disease patients is related to an impairment in the central activation of motor units. On the basis of these considerations, the reduction in Pe _max appears to reflect a reduced neural drive to expiratory muscles in our patients. Since, as already mentioned, the force developed by the contracting muscles is proportional to the corresponding IEMG activity, to appropriately compare IEMG_P values in patients and control subjects it seems reasonable to apply the correction procedure reported in Methods.

We measured the intensity of cough responses noninvasively by making use of IEMG recordings of abdominal muscles. The force developed by contracting muscles is proportional to their IEMG activity under both isometric and isotonic conditions (18, 21, 25). The rate of rise or slope of the IEMG activity reflects the rate of recruitment of motor units, as well as the increase in their firing frequency, while IEMG_P is an expression of the total number of units recruited and of their maximal firing frequency (26). Admittedly, the measurement of variables such as expiratory flow and pressures would have been useful in clarifying the dynamics of expiratory thrusts during coughing. However, we did not use a pneumotachograph because, as explained in Methods, this device might have affected the assessment of the cough threshold. On the other hand, we could not obtain reliable mouth pressure measurements during coughing due to obvious disturbances of the pressure signal induced by the cough noise. In this context, it can be recalled that in a previous study in normal subjects (18) we found that IEMG_P and IEMG_P/Tec correlate with expiratory flow. It seems conceivable that this correlation is also valid in patients with Parkinson’s disease. In fact, since there is no reason to assume that the force–IEMG relationship is disrupted in these patients, the major factor determining the rate of expiratory flow at constant airway resistance is the intensity of IEMG activity and the consequent force development by the abdominal musculature.

It can be observed that patients’ slower rate of rise of abdominal IEMG, leading to a delayed development of a positive subglottic pressure during reflex coughing, could also be associated with a delayed glottis opening, so that a slower building up of subglottic pressure may not necessarily result in a weakened cough effort. In this connection, it should be recalled that in normal subjects the glottis is actively opened, as a rule, 0.2 s after its closure, which occurs simultaneously or just after the onset of IEMG activity (27). Thus, since in control subjects Tec was approximately 0.2 s, glottis opening did occur at peak IEMG activity. During reflex cough, Tec was abnormally prolonged in the patients (Table 3); even if glottis opening was also delayed, maximal subglottic pressure and expiratory flow should have been attained at IEMG_P. Because IEMG_P is reduced in Parkinson’s disease patients, the intensity of the corresponding expiratory thrust also should be reduced accordingly.

The reduction observed in IEMG_P and IEMG_P/Tec during cough efforts probably reflects one of the most prominent functional disturbances in patients with Parkinson’s disease, i.e., slowness of movement that is ongoing, or bradykinesia. Although Tec is similar in patients and control subjects during MVC, this finding does not necessarily mean that bradykinesia in our patients is limited to RC. Although peak amplitudes of IEMG activity in AU during both MVC and RC in patients were quite similar to those observed in patients (Figure 4B, right side), it should be recalled that this was due to factors affecting recording conditions in different experimental sessions, particularly the gain of EMG amplification (see Methods). In fact, following normalization, we were able to reveal that IEMG_P values are actually reduced in the patients, as diagrammatically shown in the left side of Figure 4. The reduction in normalized IEMG_P observed in the patients also resulted in a significant reduction in the slope of abdominal IEMG activity during MVC (Table 3). This reduction can reliably be taken as an index of bradykinesia (1, 3, 4, 28, 29). A contributing feature of bradykinesia has been clearly established: it consists of a failure to energize muscles up to a level necessary for the execution of rapid or ballistic movements (3, 4, 28, 29). Because cough can conceivably be regarded as a fast or ballistic-like motor act, the same pathophysiological mechanism could account for impaired motor cough responses in Parkinson’s disease. Studies with transcranial magnetic stimulation (28, 30) indicate that not only akinesia, i.e., failure to move or slowness to initiate a movement, but also bradykinesia, may depend, at least in part, upon a deficiency in activation of the motor cortex. Rigidity may also contribute to slowness of movement but it is not the cause (31). The finding that not only the rate of rise of abdominal IEMG activity, but also its decay is slower in patients, suggests that both the recruitment and the derecruitment of abdominal motor units (Figure 4) are affected.

Although the rate of rise of abdominal IEMG activity was reduced during RC compared with MVC both in control subject and in patients, this reduction was markedly less pronounced in the latter. This suggests that RC is relatively more impaired than voluntary cough in patients with Parkinson’s disease. The reasons for this phenomenon are at present only a matter of speculation. In this connection, it can be recalled that not only cough, but also other autonomic reflexes, such as the pharyngeal phase of swallowing (15), are impaired in Parkinson’s disease.

A detailed discussion on the complex circuitry of basal ganglia is beyond the scope of the present report. We can just emphasize that the major output of the basal ganglia arises from the inner segment of the globus pallidus and the pars reticulata of the substantia nigra; the disruption of basal ganglia function in Parkinson’s disease is mainly, although not exclusively, due to profound degeneration of the nigrostriatal dopaminergic system (see, e.g., 31, 32). Pathways to the thalamus are mainly involved in the control of voluntary movements, whereas projections to the pedunculopontine nucleus of the mesencephalic pontine tegmentum and to the pontomedullary reticular formation are possibly implicated in postural and reflex influences (1, 32).

Present results, obtained in a group of patients with prevailingly mild to moderate clinical disability, clearly demonstrate a defective motor control of cough in Parkinson’s disease; the finding of an inverse relationship between IEMG variables and the level of clinical disability (Figure 3) leads us to hypothesize that, with the progression of the disease, more marked impairments in cough motor control may result in a substantial deficit of the protective mechanisms responsible for airway clearing. We cannot even exclude that in airway responsiveness to tussigenic stimuli may also be reduced in the most advanced stages of the disease. However, present data do not provide any evidence to support this possibility. Nevertheless, even in the absence of an increase in cough threshold, a defective motor control of coughing, along with the well known swallowing disturbances (10, 15), may per se account for the high incidence of potentially fatal respiratory infections in the late stages of Parkinson’s disease.