Abstract
How do the respiratory centres of patients with chronic obstructive pulmonary disease (COPD) and hypercapnia respond to acute increases in inspiratory load? A depressed respiratory motor output has long been postulated, but studies on this issue have yielded inconsistent results, partly due to limitations of investigative techniques. Many of these limitations can be overcome by the twitch interpolation technique, which is capable of accurately quantifying the degree of diaphragmatic activation, termed the voluntary drive to breathe. The hypothesis that patients with COPD and hypercapnia compensate for an acute increase in mechanical load on the inspiratory muscles with a lower voluntary drive to breathe than is the case with normocapnic patients was tested.
Measurements were obtained in 15 patients with COPD, six of whom displayed hypercapnia and nine normocapnia. The maximum degree of diaphragmatic activation, expressed as a voluntary activation index (mean±sem), was higher in hypercapnic than in normocapnic patients (98.7±0.7 versus 94.5±0.9% (p=0.006)), as was the mean value (94.5±0.7 versus 88.5±1.9% (p=0.01)). Within-patient values of the index were also less variable in the hypercapnic patients (coefficients of variation, 3.4±0.3 versus 6.1±0.9%, p=0.01). Multiple regression analysis revealed the ratio of dynamic elastance to maximum transdiaphragmatic pressure, an index of inspiratory muscle loading, and pH as the only variables that correlated with maximum voluntary activation index (r2=0.69, p=0.02 for each variable).
Contrary to the hypothesis, it was concluded that voluntary activation of the diaphragm was greater and less variable in hypercapnic patients than normocapnic patients with severe chronic obstructive pulmonary disease during an acute increase in inspiratory mechanical load. Whether greater diaphragmatic recruitment during episodes of a severe exacerbation of chronic obstructive pulmonary disease provides a survival advantage for hypercapnic patients with chronic obstructive pulmonary disease remains to be determined.
- airway occlusion pressure
- control of breathing
- diaphragmatic function
- inspiratory muscle loading
- respiratory centre drive
- twitch interpolation
This work was supported by grants from the Veterans Administration Research Service, the American Lung Association of Metropolitan Chicago, and the Gaylord and Dorothy Donnelley Foundation; A. Topeli was supported from Hacettepe University School of Medicine and Turkish Education Foundation, Ankara, Turkey.
The presence of hypercapnia in patients with chronic obstructive pulmonary disease (COPD) augurs an ominous prognosis 1. Despite extensive research, the mechanism responsible for hypercapnia is not completely understood, limiting the approach to rational therapy. Arterial carbon dioxide tension (Pa,CO2) bears only a weak relationship with the severity of airway obstruction and ventilation/perfusion mismatch 2, suggesting that nonparenchymal factors may be operative. In stable patients with COPD, Begin and Grassino 3 found that hypercapnic patients had less inspiratory muscle strength and a greater inspiratory load than normocapnic patients. The scatter in load and strength was large, however, and each accounted for less than half of the variance in Pa,CO2. Accordingly, additional factors must contribute to the development of hypercapnia.
A subnormal respiratory motor output has long been postulated as a mechanism for the hypercapnia in patients with COPD. An inability to maintain a normal respiratory motor response during an exacerbation of COPD could lead to critical hypoventilation, respiratory failure, and death. Unfortunately, investigations have yielded inconsistent results, with reports of increased 4, decreased 5, 6, and equivalent 7, 8 respiratory motor output in hypercapnic and normocapnic patients. These conflicting findings stem, in part, from limitations of techniques used to measure respiratory motor output, such as the diaphragmatic electromyogram and airway occlusion pressure 9. The twitch interpolation technique 10 overcomes many of these limitations. By recording the increase in transdiaphragmatic pressure (Pdi) that results when phrenic nerve stimulation is superimposed on a voluntary diaphragmatic contraction, the degree of diaphragmatic activation, also termed the voluntary drive to breathe, can be quantified 10, 11.
Employing this technique in patients with asthma, Allen et al. 11 found that some patients had a reduced voluntary drive to breathe, which might predispose to ventilatory failure and fatal asthma. To date, voluntary drive to breathe has not been compared in hypercapnic and normocapnic patients with COPD. The current investigation was conducted to test the hypothesis that patients with COPD and hypercapnia compensate for acute increases in mechanical load on the inspiratory muscles with a lower voluntary drive to breathe than do normocapnic patients.
Methods
Fifteen males with severe COPD were studied (table 1⇓). Six patients had hypercapnia (Pa,CO2>47 mmHg) and nine patients had normocapnia (Pa,CO2<43 mmHg). No patient with an acute exacerbation of their disease within the preceding month was enrolled. The study was approved by the Human Studies Subcommittee of Hines Veterans Administration Hospital, and written informed consent was obtained from all patients.
Airway pressure at 0.1 seconds (P0.1) after commencing tidal inspiration against an occlusion was measured using a two-way valve (Hans Rudolph, Kansas City, MO, USA) connected to a pneumotachograph (3700B Hans Rudolph) 12. End-tidal carbon dioxide tension (PET,CO2) was measured at the mouthpiece with a capnometer (Novametrix, Wallingford, CT, USA). Oesophageal (Poes) and gastric pressures (Pga) were measured using balloon-tipped catheters; Pdi was obtained by electronic subtraction of Poes from Pga. Airway pressure was measured at the mouthpiece. Transpulmonary pressure was obtained by subtracting Poes from airway pressure. Dynamic lung elastance (Edyn) was calculated as the ratio of change in transpulmonary pressure between instants of zero flow over the change in tidal volume within the same breath 13, and the average value during a 1-min recording of resting breathing was calculated.
Phrenic nerve stimulation
Compound diaphragmatic action potentials were recorded bilaterally with two sets of surface electrodes at the level of the seventh and eighth intercostal spaces and the anterior axillary line. The signals were amplified and band passed filtered (10–1,000 Hz; Gould, Valley View, OH, USA). Bilateral phrenic nerve stimulation was performed using two magnetic stimulators (Magstim 200, Jali, Newton, MA, USA) with two sets of double 40-mm coils (D40-1183.00, Jali, Newton, MA, USA) 14. The area of optimal stimulation was located by moving the stimulating probe around the posterior border of the sternomastoid muscle at the level of the cricoid cartilage. The area where stimulation elicited the compound diaphragmatic action potentials of greatest amplitude, while the patient rested at end-expiratory lung volume, was marked and subsequent stimulations were performed at this site and at 100% of the stimulators' output (magnetic field of 3.2 Tesla).
Study protocol
Patients came to the laboratory for two visits, 1 to 3 days apart. On the first visit, arterial blood gases and lung volumes were measured. On the second visit, after placement of transducers and while seated with the back supported at a 90° angle, the patient performed 10–15 Mueller manoeuvres, with ≥1 min of rest separating each manoeuvre 11. The patient was exhorted to maintain the maximal pressure for ∼2 s and oscilloscope recordings of Poes provided visual feedback 11. Maximal transdiaphragmatic pressure (Pdi,max) was measured as the best of the manoeuvres. A single phrenic nerve stimulation was delivered during each Mueller manoeuvre (superimposed-Pdi,tw), and just after completion of the manoeuvres while the patient was relaxing at end-expiratory lung volume (control-Pdi,tw).
After the last stimulation the patient rested for 1 h, and resting breathing was then recorded over 2 min. The patient then underwent a carbon dioxide rebreathing test following standard technique 15. During resting breathing and carbon dioxide (CO2) rebreathing, P0.1 was recorded every 10–20 s 5. After a rest period of ≥20 min, CO2 rebreathing was repeated.
Signal processing and data analysis
Data were recorded and digitized at 2,000 Hz. The amplitude of the superimposed-Pdi,tw was measured as the increment in Pdi produced by the stimulation delivered during the maximal effort, i.e. the difference between the maximum displacement in Pdi secondary to phrenic nerve stimulation and the Pdi value immediately preceding the stimulation. The amplitude of the control-Pdi,tw was measured as the difference between the maximum pressure displacement secondary to phrenic nerve stimulation and the Pdi value immediately preceding stimulation while the patient was relaxing at end-expiratory lung volume. Individual twitch responses were rejected from analysis according to previously described criteria (table 2⇓) 16.
The voluntary activation index of the diaphragm, which reflects the extent of diaphragmatic activation during any given inspiratory effort, was calculated 11 as:Statistical analysis was performed and unless stated otherwise, data are reported as mean±sem. Paired t-tests were used for data, and linear regression and multiple regression analysis were performed where appropriate; a two-tailed p-value <0.05 was considered statistically significant.
Results
Age, body mass, spirometry and lung volumes were similar in the two groups of patients (table 1⇑). The hypercapnic patients had a lower arterial oxygen tension (Pa,O2) (p=0.01) and, as expected, they also had a lower pH (p<0.01), and a higher Pa,CO2 (p<0.001) and arterial bicarbonate content (p<0.001).
Respiratory motor output during resting breathing, as reflected by P0.1, was similar in the hypercapnic and normocapnic patients (3.0±0.6 and 3.3±0.3 cmH2O, respectively (p=0.65; fig. 1⇓)). Responsiveness of the respiratory centres to chemical challenge, quantified as the slope of P0.1 plotted against PET,CO2, did not differ between the hypercapnic and normocapnic patients (0.31±0.12 and 0.54±0.19 cmH2O·mmHg−1, respectively (p=0.40; fig. 1⇓)). Similarly, when respiratory centre reserve was quantified as the ratio of resting P0.1 to P0.1 recorded at the end of the CO2 rebreathing manoeuvre (modified from the study of Montgomery et al. 17), there was no difference between the hypercapnic and normocapnic patients (0.42±0.09 and 0.35±0.05, respectively (p=0.46)).
Figure 2⇓ shows recordings with the twitch interpolation technique in a representative hypercapnic and normocapnic patient, with the former exhibiting complete diaphragmatic activation and the latter incomplete activation. All but one of the hypercapnic patients achieved a voluntary activation index >98% during at least one effort, whereas every normocapnic patient had values <98% (fig. 3⇓). The maximum value of the voluntary activation index was higher in the hypercapnic patients than in the normocapnic patients (98.7±0.7 and 94.5±0.9%, respectively (p=0.006)), as was the mean value (94.5±0.7 and 88.5±1.9%, respectively (p=0.01; fig. 3⇓)). Moreover, within-patient values of the index were less variable in the hypercapnic group than in the normocapnic group (coefficients of variation, 3.4±0.3 and 6.1±0.9%, respectively (p=0.01; fig. 3⇓)).
The hypercapnic and normocapnic patients exhibited equivalent diaphragmatic contractility, assessed by Pdi,max (85.4±20.1 and 98.9±10.3 cmH2O, respectively (p=0.52)) and Pdi,tw (18.6±3.2 and 20.7±3.0 cmH2O, respectively (p=0.66)). Edyn was higher in hypercapnic than in normocapnic patients (9.62±1.66 and 4.37±0.70 cmH2O·L−1, respectively (p=0.02)). The ratio of Edyn to Pdi,max, an index of inspiratory muscle load, was higher in hypercapnic than in normocapnic patients (0.14±0.03 and 0.05±0.01·L−1, respectively (p=0.03)); this ratio was correlated with Pa,CO2 (r=0.69, p=0.004) and maximum-voluntary activation index (r=0.69, p=0.004; fig. 4⇓). Maximum-voluntary activation index was also correlated with Pa,CO2 (r=0.66, p=0.007) and pH (r=−0.71, p=0.003; fig. 4⇓). When multiple regression analysis was performed using pH, Pa,CO2 and Edyn/Pdi,max as three independent variables and the maximum-voluntary activation index as the dependent variable, pH and Edyn/Pdi,max were the only factors that correlated with maximum-voluntary activation index (maximum-voluntary activation index=618.6+(21.6×Edyn/Pdi,max)−(70.8×pH) with p=0.02 for each variable and r2=0.69 for the model).
Discussion
Patients with COPD who were hypercapnic achieved almost complete activation of their diaphragm during maximal voluntary efforts, whereas patients with normocapnia displayed lower and more variable activation.
The respiratory motor output response to an acute increase in ventilatory demand, assessed by calculating the ratio of resting P0.1 to P0.1 recorded at the end of CO2 rebreathing, was of similar magnitude in the two patient groups, with considerable overlap between the normocapnic and hypercapnic patients. These data suggest that respiratory centre reserve is truly equivalent in hypercapnic and normocapnic patients or that the ratio of resting P0.1 to P0.1 at the end of a hypercapnic challenge is not sufficiently sensitive for identifying a difference in reserve.
Another similar approach for assessing the response of respiratory motor output to an acute increase in ventilatory demand is to measure the response of P0.1 to CO2 during a rebreathing manoeuvre. Most investigators have reported that the slope of this response is depressed in hypercapnic patients 6, but some 18, like the present authors (fig. 1⇑), found slopes equivalent to those in normocapnic patients. A serious limitation of this index of respiratory drive is that airway pressure can substantially underestimate intrathoracic pressure in patients with COPD because the time constant of the respiratory system is increased. The problem is aggravated during rebreathing by the progressive rise in lung volume, causing a decrease in inspiratory muscle length, and by positive recoil of the respiratory system at end-expiration 9. A more fundamental problem is that a patient's slope can cross the normal response line 19; data points that lie to the left of the intersection imply that drive is supranormal, while points to the right imply blunted drive.
The neural drive to the diaphragm during loading can be assessed using the twitch interpolation technique 10, 11, 20, 21. In rabbits, Ferguson 10 found that measurements of the degree of diaphragmatic activation, quantified with the twitch interpolation technique, were closely correlated with recordings of the phrenic nerve electroneurogram (r2=0.80–0.96). The twitch interpolation technique is extremely sensitive, and when executed carefully, can detect as little as a 1% change in twitch pressure 22. An additional attraction is that the values of diaphragmatic activation are independent of change in lung volume above functional residual capacity 23. Accordingly, the measurement diaphragmatic activation, in terms of the voluntary activation index, has several advantages over more commonly used indices of respiratory drive, including P0.1 and its response to CO2 rebreathing.
In 10 healthy subjects, Allen et al. 11 found that the mean±sd value of the voluntary activation index of the diaphragm was 88±12%. The mean value of the index in the present study's normocapnic patients (88.5±1.9%) was equivalent to that in healthy subjects, but hypercapnic patients had a higher value (94.5±0.7%). Values during several maximal efforts in the well-motivated individuals of the study of Allen et al. 11 displayed considerable within-subject scatter (range of values of the index, 40–100%). The within-subject scatter of values was less in the present study's normocapnic patients (range, 71.3–97.9%) than that in the healthy subjects of the study of Allen et al. 11, and smaller again in the present study's hypercapnic patients (range, 85.4–100%; p=0.01). The smaller scatter in the more severely ill patients (with hypercapnia and an increased mechanical load) is similar to the observed decrease in natural variability of heart rate and breathing pattern in disease states 24, 25.
Several variables modulate the extent of voluntary activation of the diaphragm. Pain arising in muscle or skin can decrease voluntary activation 26, but there is no reason to suspect this factor in the present study's patients. A low lung volume depresses voluntary activation 23, but lung volumes were equivalent in the present patient groups (table 1⇑). Hypoxaemia modulates respiratory drive 9, 27 and, although the hypercapnic patients were more hypoxaemic, no significant correlation was observed between the voluntary activation index and Pa,O2. The voluntary activation index was positively correlated with Pa,CO2 and negatively correlated with pH (fig. 4⇑). Likewise, voluntary activation index has been reported to rise with carbon dioxide tension (PCO2) during inspiratory resistive loading 28. These observations suggest that chronic respiratory acidosis may enable hypercapnic patients with COPD to generate an increased voluntary activation of the diaphragm.
That both voluntary activation index and Pa,CO2 were positively correlated with Edyn/Pdi,max (fig. 4⇑) suggests that the ability of hypercapnic patients to more fully recruit the diaphragm during acute loading represents an adaptive response to a chronically increased mechanical load. Evidence exists to support the observation that respiratory drive is not impaired in hypercapnic patients. In patients with COPD, the highest diaphragmatic motor discharge frequencies during resting breathing appeared to occur in patients with the highest Pa,CO2 levels, although the authors did not set out to determine if the discharge frequencies differed between hypercapnic and normocapnic patients 29. In healthy subjects sustaining an inspiratory resistive load, McKenzie et al. 28 noted that the development of hypercapnia was accompanied by an increase in the voluntary activation index of the diaphragm.
It is not clear what advantage, if any, increased voluntary activation of the diaphragm holds for hypercapnic patients with COPD. When faced with an elevated mechanical load, increased muscle activation predisposes to diaphragmatic muscle injury 30. Downregulation of respiratory motoneuronal output has been postulated as a useful adaptive strategy to avoid structural injury in the face of an elevated load 9, 13, and such disfacilitation has even been termed “central wisdom”. Patients with COPD who are incapable of increasing neural drive during a deterioration of their disease, however, are prone to hypoventilation and lethal hypoxaemia. In a recent study of patients with COPD who failed a trial of weaning from mechanical ventilation it was found that patients developed a >40% increase in inspiratory muscle effort as they developed progressive respiratory failure 13. That is, when faced with an acute increase in mechanical load, patients with COPD do not reduce inspiratory muscle activation, despite the theoretical advantage of doing so. Nevertheless, humans subjects undergoing extreme loading may find it increasingly difficult to maintain full voluntary activation of their muscles as progressive fatigue develops, i.e. they experience central fatigue 21. Studies in animals 27, 31 also suggest that extreme loading results in depression of drive to the inspiratory muscles.
While it is difficult to separate the contributions of voluntary and involuntary drive to the respiratory motoneurones during external loading, some evidence suggests that the bulbopontine respiratory centres do not fully activate the diaphragm even when the chemical drive to breathe is high 27, 32. The human motor cortex projects directly and extensively to the diaphragm 33 and is implicated in the production of voluntary inspiratory efforts 34. Thus, as Allen et al. 11 have noted, voluntary drive to the diaphragm may be an important defense against the development of hypoventilation during extreme respiratory loading, as can occur with an exacerbation of COPD. If some patients experiencing a severe exacerbation of COPD develop progressive central fatigue, those in whom the voluntary activation index is already low will be at a disadvantage compared with patients having a higher pre-existing voluntary activation index, as the inspiratory pressure swings in the former case might rapidly decrease to levels not sufficient for sustaining alveolar ventilation. This situation is analogous to patients with prior poliomyelitis who exhibited decreased limb muscle endurance during a fatigue protocol compared to healthy subjects, partly due to a greater degree of central fatigue, which, in turn, appears to result from an impairment in voluntary activation of the muscles 35. In other words, patients with a higher load on their respiratory muscles may have learned how to fully activate their diaphragm on an intermittent basis. This ability to achieve full activation of the diaphragm is likely to provide a survival advantage for the patient with COPD who is faced with an increased mechanical load and worsening gas exchange.
Indirect lines of evidence suggest that the extent of voluntary activation of the diaphragm that patients with COPD can achieve, while clinically stable, could influence their capacity to maintain alveolar ventilation during an exacerbation. Firstly, an increase in Pa,CO2 during an exacerbation of COPD has been found to be greater in patients who were normocapnic when clinically stable than in patients who were hypercapnic when clinically stable 36. This observation suggests that the respiratory controller recruits the inspiratory muscles to a lesser extent in patients who were normocapnic than in those who were hypercapnic (when both were clinically stable). Secondly, hypoxaemia at the time of hospital admission for an exacerbation was less severe in patients who were normocapnic when clinically stable than in patients who were hypercapnic when clinically stable 36. This observation supports the notion that gas exchange is more severely compromised in patients who are hypercapnic than in those who are normocapnic. Accordingly, when patients with long-standing hypercapnia experience an acute exacerbation of COPD, their only option is to maximally recruit their inspiratory muscles, or to face life-threatening hypoxaemia. Thirdly, mortality secondary to an exacerbation is not linked to the absolute level of hypercapnia but to the degree of acidosis 37, 38. These observations suggest that the inability to maintain an adequate alveolar ventilation by almost complete respiratory muscle recruitment may contribute to mortality.
In conclusion, during an acute increase in inspiratory load, patients with chronic obstructive pulmonary disease and hypercapnia were able to activate the diaphragm more fully than patients with the disease and normocapnia. Whether greater diaphragmatic recruitment during severe exacerbations of chronic obstructive pulmonary disease provides a survival advantage remains to be determined.
Acknowledgments
The authors gratefully thank S.C. Gandevia for helpful comments about the manuscript.
- Received February 17, 2000.
- Accepted February 7, 2001.
- © ERS Journals Ltd