INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Cough is a defense reflex mechanism initiated by appropriate chemical and/or mechanical stimulation of sensory endings located in the respiratory tract, namely those in the larynx and the proximal portion of the tracheobronchial tree (1). The main purpose of this reflex is protecting the respiratory apparatus from aspiration of foreign bodies and removing excessive airway secretions. Voluntary coughing has the same motor features as reflex cough, and can effectively be used in airway clearing.

The main source of laryngeal afferent fibers is the internal branch of the superior laryngeal nerve (2, 3). By recording from the peripheral cut end of this nerve in animals breathing through their upper airways, Sant'Ambrogio and coworkers (4) identified afferent endings activated by the inspiratory cooling of the laryngeal lumen ("flow" receptors), by changes in laryngeal transmural pressure (pressure receptors), as well as "drive" receptors, i.e., those stimulated by the contracting laryngeal muscles and the tracheal tug. In addition to these afferent modalities closely related to the events of the breathing cycle, another type of laryngeal receptors, scarcely active during normal respiration but promptly excited by various mechanical and chemical stimuli, including water solutions lacking the chloride ions, has also been described (5, 6). Tussigenic stimuli have been shown to activate the "irritant" receptors located in the laryngeal mucosa as well as the rapidly adapting "irritant" receptors of the tracheobronchial tree. Both these receptor types seem to be mainly responsible for reflex cough responses (5).

Although it is well known that cough can also be elicited from the larynx with some distinctive features (7), several recent findings have cast doubts concerning the actual role of the larynx in this airway protective reflex (3). Coughing has been found to survive laryngeal deafferentation, both in humans (8) and in animal preparations (9, 10), thus indicating that signals from the larynx do not play a necessary role in the cough reflex. In particular, in healthy subjects, anesthetic block of the superior laryngeal nerves (SLNs) abolishes cough elicited by probing of the laryngeal mucosa, but does not significantly affect the response to citric acid inhalation, thus suggesting an involvement of the tracheobronchial tree in coughing elicited by citric acid stimulation. Noticeably, in heart-lung transplanted patients with denervated lungs but intact SLNs, the cough response elicited by inhalation of nebulized water is inhibited, whereas coughing in response to direct stimulation of the laryngeal surface by direct instillation of water droplets remains intact (11).

These findings seem to indicate that the larynx is not adequately stimulated by inhalation of fog, and that vagal receptors located in the tracheobronchial tree may play a major role in the mediation of cough responses to this stimulus. However, these findings can also be explained by taking into account the lack of pulmonary stretch receptor influences, which are known to facilitate expiratory motoneurons discharge (12). The role of sensory feedback from the larynx in the cough reflex has also been questioned. For instance, the pattern of activation of the intrinsic laryngeal muscles during coughing elicited by tracheal probing is unaffected by section of the SLNs, suggesting an entirely automatic, preprogrammed nature of this event (13). All the aforementioned reported issues underline the need of further experimental data to clarify the role of laryngeal receptors in cough generation.

The motor pattern of coughing, which aims at generating the high flow velocities during the expulsive phase involves several groups of muscles, all having a prevalent respiratory function. These muscles are brought into action sequentially, with a well-coordinated and fixed pattern which basically consists of a deep inspiration followed by a strong expiratory effort, initially against a closed larynx, which then suddenly opens releasing a strong blast of air through the airways (14). Although evidence exists (15) that tracheostomized and intubated patients still "cough," suggesting that airway clearing may still be possible even in the absence of glottis closure, laryngectomy may limit cough effectiveness and represents an indication to bronchial hygiene therapy, particularly directed cough (16). In addition, it is not clear whether ablation of the larynx has an impact on respiratory muscle activity during voluntary and reflex expiratory thrusts. It has been reported that signals conveyed by the SLNs may have a role in regulating the strength of the inspiratory and expiratory efforts during coughing (7, 17, 18). Thus, it can be expected that ablation of the larynx implies not only the loss of the protective shield represented by the nose and upper respiratory tract, but also a possible interference with the physiological mechanisms implicated in cough sensitivity and/or in the control of the intensity of expiratory muscle activity during coughing.

The present experiments were undertaken to investigate the impact of surgical ablation of the larynx on the sensory and motor component of coughing either voluntarily produced or reflexly evoked by inhalation of ultrasonically nebulized distilled water (fog). To this end, we evaluated the threshold for the cough reflex, as well as the intensity of both voluntary and reflex cough efforts, in patients laryngectomized for cancer of the larynx and in age-matched control subjects.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subjects

Ten male patients age 46 to 76 yr (mean 67.3) participated in the study. All patients had total removal of the larynx for laryngeal carcinoma between 24 and 72 mo previously. Laryngectomy was always performed according to the well-established, classic approach consisting of the complete larynx exposure, total excision of the organ, section of laryngeal pedicles (including the recurrent and superior laryngeal nerves), and hypopharyngeal reconstruction. The vagal cranial nerve was spared bilaterally, and patients were left with a definitive tracheostoma through which they performed all respiratory functions. At the time of the study, they were fully active and fit; none had evidence of local or metastatic recurrence of the carcinoma. All patients reported a previous smoking history; however, they had refrained from smoking since the disease had been diagnosed. Ten healthy, age-matched, lifetime nonsmoker volunteers (8 males and 2 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 and was approved by the local ethical committee. 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 (19). Respiratory function data of both patients and control subjects are summarized in Table 1.

Experimental Procedures and Protocol

Most of the experimental procedures have already been fully described (20, 21). During cough challenges, all participants breathed through a 2-way, low-resistance shutter valve (modified Lloyd valve; Warren E. Collins Inc., Braintree, MA) with a dead space of 46 ml. The expiratory port of the valve was connected to a number 4 Fleisch pneumotachograph and a flow transducer (HP47304; Hewlett-Packard, Palo Alto, CA); the inspiratory port of the valve was in series with the ultrasonic nebulizer via a large-bore plastic tubing.

The accuracy of the pneumotachograph-transducer assembly had preliminarily been assessed, in ATPS conditions and with all adapters used for connecting the instruments to the airway opening of both patients and control subjects, by using a calibration device similar to that developed by Petusevsky and coworkers (22) and capable of generating high flow rates (range 0 to 7.5 L/s) by explosive decompression of air contained in a metal cylinder. This calibrating device was alternatively connected to the Fleisch pneumotachograph and the reference spirometer (Collins Survey II, Warren E. Collins Inc., Braintree, MA). Acquisition of volume and flow signals was always performed by means of a personal computer after analog-to-digital conversion at 5,000 Hz. The pneumotachograph flow signals proved to closely correspond (± 1.5%) to those derived from the volume-time curves obtained with the Stead-Wells spirometer.

Reflex cough (RC) was induced by inhalation of fog produced by a 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 (20). The range of nebulizer outputs employed in the present study was from 0.08 to 4.45 ml/min. Maximal voluntary cough (MVC) efforts were obtained by repeatedly encouraging each participant to cough as forcefully as possible as in a strong attempt at clearing of the airways. The lung volume at which these expulsive efforts were commenced was not controlled.

The force of expiratory muscles was measured by means of a portable pressure transducer (Spirovis, Cosmed, Italy) as the maximal static expiratory pressure (PE_max), i.e., the highest pressure generated by a subject against the closed airway and sustained for at least 1 s after a full inhalation to near total lung capacity (TLC) (23). Control subjects were connected to the transducer by wearing a mouthpiece in the usual way with a noseclip; during these measurements their cheeks and floor of the mouth were supported with the palm of the hands by an investigator. In the patient group, the teeth lugs were cut off the mouthpiece, which was attached over the tracheostomy stoma with airtight sticky plaster; the measuring apparatus was also gently but firmly pressed against patients' neck to ensure seal between the tracheostomy stoma and the transducer. In all instances, participants were 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 (24). 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 also fed to a DC amplifier whose gain could be adjusted to obtain recordings of such an amplitude to allow accurate measurements. Before each challenge participants were asked to change their posture (trunk flexion), and the IEMG wave forms evoked by these maneuvers were compared with those recorded during voluntary coughing for differentiation. To the same end, control subjects were also asked to simulate events such as exhalation of a long, audible breath and throat-clearing.

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 (25), the back of the chair was tilted backward by approximately 30 degrees. In these conditions, abdominal muscles displayed no obvious rhythmic or tonic activity. Participants were then requested to perform 8 to 10 MVC maneuvers, each separated by a 5- to 10-s interval, during which both the abdominal IEMG activity and the expiratory flow were recorded. Three to five PE_max maneuvers were also performed. In the patients, leaks were excluded by a carbon dioxide analyzer probe seeking escaping CO₂. During cough trials, patients and control subjects were connected to the breathing apparatus with the same technique as that employed for PE_max measurements. After a 10-min recovery period, each subject inhaled during resting tidal breathing increasing fog concentrations obtained by adjusting the nebulizer output. In all instances, the fog inhalation time was standardized at 1 min for each nebulizer output; 2 to 3 min of rest were introduced between steps. 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 (20). After a further 30-min recovery period, suprathreshold (ST) cough responses were evaluated by having all participants inhale, for 1 min, the fog concentration corresponding to approximately 1.8 times the previously established threshold (T) fog concentration.

We were concerned that the use of the Lloyd valve could affect the characteristics of the aerosol particles produced by the nebulizer, and, hence, the assessment of cough threshold (20). To investigate this possibility, we preliminarily compared the results of double assessments of cough threshold, i.e., with and without the Lloyd's valve, in a group of normal volunteers (n = 15). It was found that median cough threshold values obtained with and without the valve were similar (1.31 ml/ min in both instances; p > 0.05, paired Wilcoxon test).

Recordings of the studied variables were performed by means of an eight-channel chart recorder (HP 7758A; Hewlett-Packard) with a paper speed of 5 to 25 mm/s.

Data Collection and Analysis

We measured the peak of IEMG activity (IEMGP) in arbitrary units during PE_max, MVC, during each of the two fog challenges required to determine reflex cough at threshold level (RCT), as well as during reflex cough efforts at suprathreshold stimulus intensity (RCST). Measurements of IEMG amplitudes recorded in different experimental session cannot reliably be used for within- and between-subject comparisons without adequate processing because they are affected by several factors such as muscle size (which may considerably vary with the gender and age of the subject), the efficiency of skin-electrode coupling, skin resistance, electrodes position, distance between electrodes, and adjustment of signal amplification (21). To overcome this problem, all IEMGP values recorded in each participant during PE_max maneuvers, MVC, and RC were expressed, in accordance with common normalization procedures, as a fraction of the highest IEMGP value recorded throughout each experimental session. The highest IEMGP value was consistently attained during MVC. The normalized IEMGP values, expressed as relative units (RU), were subsequently used for all analyses.

The duration of the rising phase of the abdominal IEMG activity during both MVC and RC was measured as the time elapsed between the onset of IEMG activity and the peak level of that activity. This time duration, i.e., the time duration of the expiratory IEMG ramp during cough, was termed TEC. The onset of IEMG activity was arbitrarily considered as the time at which the activity just exceeded the 10% of its peak amplitude above the mean level of ongoing baseline activity. The 10% level was chosen to avoid uncertainties in the measurements, especially when slow drifts in IEMG activity were present. The ratio of IEMGP to TEC (IEMGP/TEC) was subsequently calculated for all considered cough efforts. Because of the proportionality between force and IEMGP and, as a consequence, between the rate of rise of the force generated by the contracting muscles and their IEMGP/TEC, these IEMG-related variables were confidently used as indices of cough intensity (20, 21). In each subject, the static expiratory maneuver showing the highest PE_max value was selected for analysis. The selected PE_max consistently displayed, compared with all other static expiratory efforts, also the highest IEMGP value. In the results, PE_max was expressed as a percentage of subject's predicted value (23). As for the MVC, the three maneuvers with the highest IEMGP were considered in each subject. Differences in these IEMGP values during MVC were always less than 10% of the maximum. For RC, all efforts recorded during cough threshold assessment were analyzed. Owing to the small variations in IEMG variables during both MVC and RC in each subject, average values were taken as single measurements for purpose of analysis.

In two subgroups of randomly selected participants (n = 8, four patients and four age-matched male control subjects), the flow signals generated during voluntary and reflex cough efforts were analyzed on-line by means of a personal computer after analog-to-digital conversion at 1,000 Hz. For each considered cough effort, both the rate of cough peak flow and the time elapsed from the onset of cough to cough peak flow (26) were measured; the ratio of cough peak flow to the time to peak, i.e., the volume acceleration (26, 28), was subsequently calculated.

Comparisons between cough threshold values observed in patients and control subjects were performed by the unpaired Wilcoxon test. Flow- and IEMG-related variables during PE_max maneuvers as well as during reflex and maximal voluntary cough efforts were analyzed by using nonparametric analysis of variance (ANOVA) followed by the Dunn's multiple comparisons tests. Linear regression analysis was used to investigate the relationships between IEMGP, IEMGP/TEC, and cough peak flow during MVC, RCT, and RCST. Reported data are means ± SD, unless otherwise stated. In all instances, p < 0.05 was taken as significant.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

All participants coughed in response to fog inhalation. Cough threshold values ranged from 0.40 to 3.26 ml/min both in control subjects and patients; median cough threshold values were 1.31 and 0.73 ml/min, respectively. The corresponding first and third quartiles were 0.73 and 2.29 ml/min in control subjects, and 0.40 and 1.62 ml/min in the patients, with interquartile ranges of 1.55 and 1.22 ml/min, respectively. Despite lower cough threshold values in the patients, the difference between the two study groups failed to reach statistical significance.

Patients' PE_max, expressed as a percentage of predicted values, was 101.76 ± 20.53 (Table 1); this value did not significantly differ from that of control subjects (99.50 ± 9.87). As reported in Table 2, IEMGP values attained by control subjects during PE_max, MVC, RCT, and RCST did not significantly differ; in contrast, patients' IEMGP values recorded during RCT consistently displayed lower values (p < 0.01) than those observed during PE_max, MVC, and RCST. In addition, during RCT, patients' IEMGP values were significantly lower than those of control subjects (p < 0.05). On the other hand, TEC values recorded in both groups during MVC, RCT, and RCST were similar. In control subjects, owing to the absence of significant differences in IEMGP and TEC values, also the rate of rise of IEMG activity (IEMGP/TEC) during both reflex and voluntary cough efforts was similar. In the patients, IEMGP, TEC, and, as a consequence, IEMGP/TEC values during MVC and RCST were also similar and not significantly different from those observed in control subjects under the same experimental conditions; however, because of the reduction of IEMGP, the rate of rise of patients' abdominal IEMG activity during RCT turned out to be significantly lower than that observed during MVC and RCST, as well as that observed during RCT in control subjects (Table 2). Average patterns of abdominal IEMG activity during MVC, RCT, and RCST are diagrammatically illustrated in Figure 1 for both patients and control subjects.

View larger version (12K): [in this window] [in a new window]   Figure 1.   Diagrammatic representation of average IEMG patterns during voluntary (solid lines) and reflex cough at threshold stimulus intensity (dotted lines) and at suprathreshold (1.8 × T) stimulus intensity (dashed lines) in control subjects (A) and laryngectomized patients (B). Reported values are means ± SD for peak IEMG activities (in relative units, RU) and duration of expiratory IEMG ramp activity. Horizontal bars represent standard deviations for mean duration of expiratory IEMG ramp activity; standard deviations for mean peak IEMG amplitudes are represented by vertical bars. The rate of rise of IEMG activity is represented by the slope of oblique lines. *p < 0.05 for peak IEMG activity compared with control subjects.

In all instances, peak flow rate during coughing consistently displayed a behavior similar to that of IEMG-related variables. This phenomenon, which is confirmed by regression analysis, is illustrated in Figure 2 by means of representative, original recordings obtained in a control subject (Figure 2, left panels) and in a laryngectomized patient (Figure 2, right panels). As reported in Table 2, controls' cough peak flow attained comparable levels during MVC as well as during RCT and RCST; in the patients, this variable consistently displayed lower values than in control subjects during RCT, but not during MVC and RCST. The scrutiny of results revealed that the time elapsed from the onset of a cough effort, either voluntarily or reflexly produced, and the peak of expiratory flow (i.e., the time to peak) was longer in the patients (see, e.g., Figure 2). To further analyze this aspect, a computer analysis of the flow signal during voluntary and reflex coughing was performed in two subgroups of randomly selected patients (n = 4) and control subjects (n = 4). In these subgroups, the mean values of cough expiratory flow recorded during MVC, RCT, and RCST were in the same range as those reported in Table 2 for the whole populations studied. However, in the subgroup of patients, mean values of time to peak (Table 3) were significantly longer than those of control subjects (p always < 0.01). Thus, patients' cough volume acceleration values (i.e., the ratio of cough peak flow to the corresponding time to peak) obtained during MVC, RCT, and RCST turned out to be consistently lower than those observed in the subgroup of control subjects in the same experimental conditions (Table 3 and Figure 3). ANOVA revealed that in control subjects volume acceleration values were similar in all experimental conditions, whereas in the patients volume acceleration values attained during RCT were significantly lower than those recorded during both MVC and RCST. Patients' lower volume acceleration values proved to be mostly due to a significant (p always < 0.01) lengthening in times to peak; however, during RCT, lower cough peak flow values also contributed.

View larger version (13K): [in this window] [in a new window]   Figure 2.   Representative, original recordings obtained from a control subject (left panels) and a laryngectomized patient (right panels) during maximal voluntary coughing (upper panels), reflex cough at threshold stimulus intensity (middle panels), and cough at ST stimulus intensity (lower panels). Traces in each panel are cough expiratory flow and the IEMG. Compared with control subjects, patients commonly displayed a more prolonged activation of the expiratory muscles and a slower decay in expiratory flow rate during intense cough efforts either voluntarily or reflexly produced.

View larger version (16K): [in this window] [in a new window]   Figure 3.   Diagram showing the ratio of mean cough peak flow versus mean time to peak during MVC (solid lines), RCT (dashed lines) and RCST (dotted lines) levels in four control subjects (open circles) and four laryngectomized patients (filled circles). The slope of each line is the cough volume acceleration. Steeper lines correspond to higher volume acceleration values.

In control subjects, both the peak and rate of rise of abdominal IEMG activity correlated well with cough peak flow not only during MVC, but also during RCT, and RCST; similar although weaker correlations were found in laryngectomized patients under the same experimental conditions. In control subjects, the correlation coefficients of IEMGP versus cough peak flow during MVC, RCT, and RCST were 0.83, 0.84, and 0.74, respectively (p always < 0.01); in the patients, these correlation coefficients were 0.66, 0.73, and 0.66, respectively (p always < 0.01). The correlation coefficients of IEMGP/TEC versus cough peak flow during MVC, RCT, and RCST in control subjects were 0.70, 0.90, and 0.71, respectively (p always < 0.01); in the patients they were 0.59 (p < 0.01), 0.57 (p < 0.05), and 0.68 (p < 0.01).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The most important findings of the present study are that cough threshold to fog inhalation in patients with a permanent tracheostomy, although lower, does not significantly differ from that observed in age-matched control subjects, whereas the intensity of reflex motor responses, as indexed not only by the peak and rate of rise abdominal IEMG activity (20, 21), but also by the cough peak flow (26, 28), is significantly depressed in the patients at threshold stimulus strengths. Finally, in all experimental conditions, the intensity of abdominal muscle contraction closely correlates with the corresponding expiratory flow rate.

The observation that cough threshold to fog inhalation is not influenced by surgical ablation of the larynx is in keeping with the results of a previous study showing that, in normal subjects, bilateral SLN blockade by lidocaine abolished coughing elicited by probing of the laryngeal mucosa, but did not significantly affect cough threshold to citric acid (8). This result provides additional evidence that receptors possibly located in the tracheobronchial tree and innervated by the vagus nerve, are involved in the mediation of cough responses evoked by inhaled tussigenic agents even in the absence of afferent information from the larynx (8).

The pivotal role played by lower airway receptors on chemically induced cough responses is also indicated by the recent finding that coughing in response to fog inhalation is abolished in almost all heart-lung transplant patients, i.e., in patients with intact SLNs but denervated lungs (11). This phenomenon may depend upon the loss of signals from receptors known to be highly responsive to tussigenic stimuli, such as the rapidly adapting "irritant" receptors (1, 5), as well as upon the lack of the permissive role for the cough reflex of signals originating from the slowly adapting "stretch" receptors (12). The finding of a normal cough threshold in laryngectomized patients would suggest that information conveyed by the SLN is neither necessary for cough elicitation nor significantly affecting the stimulus intensity required to evoke a cough response during inhalation challenges.

This possibility, however, needs to be taken with caution for at least two reasons. First, in order to compare both the intensity of expiratory muscle contractions and the flow rates during voluntary and fog-induced cough efforts between control subjects and laryngectomized patients, the former were preliminarily screened for being cough-responders to the stimulus used. In consequence, cough threshold in our control group may have been somewhat lower than that of the actual normal population, which also comprises subjects (approximately 15%) who do not cough even when exposed to the maximal attainable fog concentration (20). This might suggest that, when compared with that of the general normal population, cough threshold to fog inhalation is actually lower in laryngectomized patients. On the other hand, we cannot even rule out the possibility that some fog-insensitive subjects are included in the general population of laryngectomized patients. Second, all our control subjects were lifetime nonsmokers, as opposed to the laryngectomized patients who had a past history of heavy smoking. Because matching the patients with control subjects with a comparable past smoking history (see METHODS) proved to be difficult, the choice of nonsmokers as a control group appears to be reasonable. The consequences of our choice are difficult to evaluate; in fact, no information is available, to our knowledge, regarding differences in cough sensitivity between ex-smokers and nonsmoking subjects.

Finally, it remains to be established whether present observations on fog-induced cough efforts can also be extended to patients with different respiratory diseases, or to cough responses evoked by different tussigenic agents such as, for instance, citric acid and capsaicin. This latter agent, as well as inflammatory mediators released during respiratory disease, may also stimulate sensory pathways other than those believed to be implicated in cough evoked by inhalation of water solutions (6, 29).

The results also indicate that, during cough evoked by threshold stimuli, the intensity of expiratory muscle contraction, when indexed in terms of both the peak (IEMGP) and rate of rise (IEMGP/TEC) of the integrated IEMG activity (20, 21), is reduced in laryngectomized patients, despite normal expiratory muscle force as witnessed by normal PE_max values. Thus, patients' reduction in IEMGP and, as a consequence, in IEMGP/ TEC during RCT is consistent with the possibility that, in the absence of afferent information from the larynx, the mechanisms subserving motor unit recruitment and/or regulating their firing rate are at least partially impaired at cough threshold level.

Several lines of evidence seem to indicate that signals conveyed by the SLN may affect the discharge of expiratory bulbospinal neurons and, hence, the activity of the abdominal motoneurons (17, 18). In the anesthetized cat, trains of stimuli to the SLN suppress inspiratory muscle activity but elicit tonic, CO₂-dependent, abdominal muscle activity associated with the appearance of a tonic discharge in bulbospinal expiratory motoneurons (17). Interestingly, in similar animal preparations, Jodkowski and Berger (18) were able to show that cold water instillation in the larynx causes inhibition of phrenic activity and activation of internal intercostal nerve activity, indicating expiratory effort. In these latter experiments, however, the larynx was not isolated, and the possibility of simultaneous activation of different airway receptors by water cannot be ruled out. In humans, repetitive, voluntary glottal closures causing expiratory flow interruption at 5 to 8 Hz, elicit bursts of activity from the expiratory muscles during the periods of no flow (30). Noticeably, when flow interruptions with the same frequency were performed by voluntarily occluding the airway with the tongue or by means of an electrically operated shutter valve, no synchronous abdominal muscle discharges can be observed. Because all these methods of flow interruption, which can be regarded as functionally equivalent to low-intensity cough efforts, are also mechanically similar at the pulmonary and expiratory muscular level, it can be concluded that laryngeal motor acts are tightly coupled with expiratory muscle activation, possibly via a centrally programmed coordination program and/or stretch reflexes arising from muscles of the upper airway (30). Therefore, it seems conceivable that ablation of the larynx may result in a disruption of these neural mechanism, thus leading to a weakened cough response to low-intensity stimuli. However, this impairment may be overcome by a more intense stimulation of water-sensitive receptors, such as that achieved during inhalation of ST fog concentrations, possibly due to the recruitment of a larger number of pulmonary water-sensitive endings and, hence, of a larger number of afferent pathways and expiratory motor units.

The correlation we found between peak expiratory flow and IEMG-related variables during voluntary and reflex cough efforts confirms and extends previous observations obtained in this laboratory during voluntary coughing in normal volunteers (20). In addition, a combined study of flow- and IEMG-related variables provides information on both neural activation and pulmonary mechanics, thus allowing a more comprehensive evaluation of cough motor responses. It seems of interest that cough expiratory flow during MVC and RCST in the patients are only slightly and not significantly lower than in control subjects. This finding strongly suggests that, after ablation of the larynx, the lack of the compressive phase of coughing, i.e., the functional aspect that differentiates more cough from other respiratory expulsive maneuvers, has little impact on flows produced during intense expiratory efforts produced either voluntarily or in response to ST stimulation.

To the best of our knowledge, there are few studies on the physical variables of coughing (26, 28), particularly volume acceleration (26, 28). In the study by Yanagihara and coworkers (28), volume acceleration values can be derived from measurements of peak flow rates and times to peak obtained in five young normal subjects during a single "gentle" cough. Volume acceleration values of normal individuals during maximal cough efforts performed at nearly TLC have been reported by Knudson and coworkers (26). Noticeably, whereas mean volume acceleration values derived from data presented by Yanagihara and coworkers (28) are lower than those reported here, values recorded by Knudson and coworkers (26) are considerably higher than those of our control subjects. To account for these discrepancies, one should consider that in the present investigation cough inspiratory volume was not controlled and participants were simply asked to cough forcibly as in an attempt of clearing of the airways. As was apparent from the visual inspection of the subjects, these voluntary cough efforts were performed at lung volumes less than TLC, but well above a resting tidal volume. Thus, it is not surprising that, in accordance with the tension-length relationship, the force developed by the contracting muscles during coughing by our subjects was lower than that attained by the subjects studied by Knudson and coworkers (26), who were asked to cough maximally from full inspiration, and higher than that reported by Yanagihara and coworkers, in which the inspired volume was no greater than a tidal volume (28).

Noticeably, patients' volume acceleration values proved to be consistently lower than those of control subjects. During MVC and RCST the major factor responsible for this phenomenon is a lengthening in time to peak, given the similarity in cough peak flow rates (Table 3); lengthening of time to peak during maximal voluntary cough efforts in laryngectomized patients has already been reported by others (27). In contrast, during RCT, also patients' lower cough peak flow values appear to play a significant role. Although it is not clear why time to peak during coughing in laryngectomized patients is increased, two distinct mechanisms may be taken into account to explain this finding. First, in normal subjects subglottic pressure during the expulsive phase of coughing abruptly falls to a level corresponding to approximately 50% of that attained during the compressive phase of the reflex (28). This phenomenon indicates that, despite active opening of the vocal folds (31), the larynx remains a site of reduced airway cross-sectional area offering a certain degree of resistance to cough flow. Because flow velocity is proportional to the peak flow rate and inversely related to the airway cross-sectional area, even an open larynx may contribute to increase cough flow velocity and, hence, cough volume acceleration. Therefore, ablation of the larynx may result, at least in functional terms, in an increase in the airway cross-sectional area leading to a reduction in both cough flow velocity and cough volume acceleration. A reduction in volume acceleration could explain why, in the absence of the larynx, it takes longer to attain cough peak flow values similar to those of control subjects.

A second possibility deals with the lack of the compressive phase, i.e., the phase in the cough motor sequence during which the inspired gas is strongly compressed against the closed glottis, and which precedes the abrupt expiratory phase of the reflex (14). Because the tension developed by the contracting expiratory muscles, for any given level of muscle activation, is higher during isometric (glottis closure) than isotonic contractions (32), the lack of a compressive phase likely reduces expiratory muscle force and, hence, cough volume acceleration during the subsequent expiratory phase. In this connection it seems appropriate to recall that the sustained activation of the expiratory muscles which characterizes MVC and RCST of laryngectomized patients (Figure 2) suggests engagement of expiratory muscles in an expulsive effort of longer duration, although of less marked intensity, which, in turn, may represent a compensatory mechanism by which laryngectomized patients try to attain cough flow rates and time courses similar to those of control subjects. The neural mechanisms underlying this prolongation of expiratory muscle activation are at present highly speculative; however, the lack of afferent signals of laryngeal origin may conceivably be involved.

In conclusion, the findings suggest that ablation of the larynx does not significantly affect cough sensitivity to inhaled nebulized distilled water; however, the presence of an intact upper airway may be of importance in the regulation of the neural mechanisms implicated in the recruitment of motor units during the expulsive phase of coughing, especially when the intensity of the cough stimulus is weak or near threshold level. The fact that people who have had a laryngectomy can still cough effectively in response to high-intensity stimuli and voluntarily but not in response to stimuli of lower intensity is in keeping with the notion that reduced ability in expectorating secretions from the lungs, along with the loss of the protective function of the nose and upper respiratory tract, can worsen a respiratory tract infection (33). Therefore, laryngectomized patients should not only be encouraged to cough forcibly on a daily basis, but also participate in bronchial hygiene therapy programs, including directed cough (16), in order to facilitate mucus removal and reduce the risk of serious chest infections.