INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Hypercapnia often complicates severe chronic obstructive pulmonary disease (COPD), and when present is associated with a worse prognosis (1, 2). Mechanisms that contribute to hypercapnia in patients with COPD include alveolar hypoventilation and an increased physiologic dead space. Airflow obstruction, as a result of intrinsic airways disease and decreased lung elastic recoil increasing airways resistance (3), leads to air trapping and hyperinflation. Air trapping and hyperinflation in turn place the diaphragm and other inspiratory muscles at severe mechanical disadvantage (4), producing alveolar hypoventilation and hypercapnia (4, 5). An increased dead space to tidal volume ratio (VD/VT) also contributes to hypercapnia (5) in severe COPD.

Lung volume reduction surgery (LVRS) has been shown to increase lung elastic recoil and improve various physiologic and functional parameters (3, 6). The effect of LVRS on hypercapnia is variable. Some studies demonstrate a reduction in Pa_CO2 (3, 7, 10) after LVRS, while others show no change (6, 14).

Perhaps because of the poor prognosis observed in COPD patients with severe hypercapnia independent of LVRS, some investigators have proposed excluding patients with hypercapnia (Pa_CO2 > 50 to 55 mm Hg) from this surgery (7, 13, 15, 16, 18, 19). Presently, however, there are no convincing data to support this recommendation. In fact, there are no data that fully describe the response to LVRS in patients with a wide range of Pa_CO2, nor has there been a well-conducted examination of the relationship between the change in Pa_CO2 after LVRS, and various functional and physiologic parameters.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patient Selection

Our study was performed in 33 consecutive patients who were not excluded on the basis of baseline hypercapnia, enrolled in the LVRS program at Temple University Hospital. All patients had severe, stable COPD and were receiving optimal medical therapy (bronchodilators, inhaled and/or oral corticosteroids, home oxygen therapy, and comprehensive outpatient pulmonary rehabilitation). All patients had quit smoking at least 6 mo before LVRS. Inclusion and exclusion criteria for admission into the LVRS program are listed in Table 1.

Before and 3 to 6 months after LVRS, all 33 patients underwent detailed physiologic measurements including spirometry, body plethysmography, single-breath diffusion capacity, 6-min walk test, symptom-limited maximal cardiopulmonary exercise, and measurements of maximal static inspiratory and expiratory mouth pressures. All patients also underwent measurement of transdiaphragmatic pressures during maximal static inspiratory efforts and during bilateral supramaximal electrophrenic twitch stimulation. All patients survived surgery and were available for follow-up studies.

Pulmonary Function Testing

Pulmonary function testing was performed using American Thoracic Society guidelines (20) with a System 6200 Autobox DL Plethysmograph (SensorMedics Corp., Yorba Linda, CA). Only postbronchodilator values are reported. Forced vital capacity (FVC) and forced expiratory volume in one second (FEV₁) were measured. Thoracic gas volumes were measured by both body plethysmography and helium dilution. Diffusion capacity for carbon monoxide (DL_CO) was assessed by single-breath technique. Maximum voluntary ventilation (MVV) was measured with the patients seated upright and instructed to breathe maximally in a deep and rapid manner for a sustained period of 12 s.

Exercise Testing

All patients underwent incremental maximal treadmill exercise testing (Precor, 9.4 sp; Precor Inc., Brothell, WA) starting at 0% incline and 1 mph. The incline was increased by 3% and speed by 1.5 mph every 3 min until symptom-limited maximum was achieved. During the test, oxygen consumption (O₂), carbon dioxide production (VCO₂), minute ventilation (E), tidal volume (VT), and breathing frequency (f_b) were recorded by a metabolic cart (SensorMedics 2900). Transcutaneous oxygen saturation (N-200; Nellcor, Chula Vista, CA) and electrocardiogram (ECG) (ECG Horizon; SensorMedics) were continuously recorded. On a separate day, the total distance that the patient was able to ambulate in a corridor for 6 min was recorded (21).

Blood Gas Analysis

Arterial blood gas was sampled from the radial artery with the patients seated upright, 20 min after receiving a bronchodilator treatment and breathing room air.

Respiratory Mouth Pressures

Respiratory mouth pressures were measured according to the method reported by Black and Hyatt (22). Maximal inspiratory pressure (PI_max) was measured from FRC, and maximal expiratory pressure (PE_max) was measured near TLC. Tests were repeated until 3 values varied less than 5% of one another. The average of 3 tests is reported.

Transdiaphragmatic Pressure

After topical anesthesia with 4% lidocaine gel, two thin-walled, balloon-tipped catheters were placed into the nares; one into the lower esophagus (esophageal pressure, Pes) and the other into the stomach (gastric pressure, Pga) (23). Both catheters were connected to pressure transducers (range ± 100 cm H₂O; Validyne, Northridge, CA). Transdiaphragmatic pressure (Pdi) was continuously displayed as the electronic subtraction of Pes from Pga. Maximal transdiaphragmatic pressures were measured during a maximal sniff maneuver (Pdi_{max sniff}) in 15 patients (24). The average of 3 values of Pdi_{max sniff} during each separate maneuver, all within 5% of each other, is reported.

Electrophrenic Stimulation

Compound diaphragm action potentials (CDAP) were measured bilaterally by a pair of 3-mm electromyogram (EMG) surface electrodes placed 2 mm apart in the seventh intercostal space in the anterior axillary line. In each patient, the area for optimal phrenic nerve stimulation was located by using anatomic landmarks (25). Once identified, an electrical stimulus was applied and the CDAP was displayed on a recording oscilloscope to confirm phrenic nerve stimulation. An abdominal plaster cast was placed over the anterior abdomen to prevent diaphragm shortening by minimizing outward displacement of the abdominal wall during electrophrenic stimulation. Stimulus voltage was incrementally increased until there was no further increase in CDAP amplitude. Once maximal stimulus voltage was achieved, it was further increased by 20% to ensure supramaximal diaphragm activation.

A modified neck brace housing the left and right phrenic nerve stimulus probes was used to ensure consistency in phrenic nerve stimulation. The phrenic nerves were then stimulated transcutaneously (S88 Stimulator; Grass, Quincy, MA) with 100 to 140 volts (approximately 30 mA), 0.1 ms in duration, to produce diaphragm twitch pressures (Pdi_twitch). There was approximately a 20- to 30-min interval between the end of Pdi_{max sniff} maneuvers and the onset of Pdi_twitch testing.

Pdi_twitch at FRC

With the patient seated in the upright posture, bilateral phrenic nerve stimulation was delivered to 11 patients at FRC after closure of an in-line 3-way valve at end-expiration. Pes was continuously monitored to ensure that end-expiratory lung volume had returned to a consistent baseline value prior to valve closure. Six to 15 consecutive twitches (each twitch separated by at least a 3-s pause) were delivered at FRC. Twitches analyzed were those considered acceptable after ensuring that FRC was at baseline and twitch morphology was consistent. Three values all within 5% were averaged and reported as Pdi_twitch.

Surgical Technique

Lung resections were performed by the same cardiothoracic surgeon via median sternotomy and bilateral stapling. The goal for resection was to remove 20 to 40% of the volume of each lung. High-resolution computed tomography of the chest and quantitative ventilation-perfusion scans were used preoperatively to target resection of lung regions with the worst emphysema, poorest perfusion, and greatest gas trapping.

Data Analysis

All data are expressed as mean ± SD except where otherwise noted. Student paired two-tailed t tests were used to compare data before and after LVRS. Linear and multiple regression analyses were used to evaluate associations between levels of hypercapnia and lung function parameters post-LVRS. All statistical analyses were conducted using a commercially available computer software program (SigmaStat Version 1.0; Jandel Co., San Rafael, CA). A p value of < 0.05 was considered statistically significant.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patient Characteristics

The study group was comprised of 33 patients (age 57 ± 8.3 yr [mean ± SD]; 21 women). All patients were well nourished at baseline. The etiology of emphysema was tobacco- related in all patients. Only three patients were oxygen-dependent at study enrollment, and 13 patients received daily systemic steroids. All patients completed 8 wk of outpatient pulmonary rehabilitation before LVRS. Demographic data are presented in Table 2.

Physiologic Data

Table 3 shows baseline and post-LVRS spirometry, lung volume, and gas exchange data in all patients. At baseline, all patients had severe airflow obstruction, air trapping, hyperinflation, and reduced MVV and diffusion capacity for carbon monoxide corrected for alveolar volume (DL_CO/VA). Gas exchange was also abnormal (Pa_O2 68 ± 13 mm Hg and Pa_CO2 44 ± 7 mm Hg).

Three to 6 mo after LVRS, there were significant increases in FEV₁ (p = 0.0001), MVV (p = 0.0001), and a trend toward an increase in DL_CO/VA (p = 0.08). There were also significant reductions in the degree of hyperinflation (TLC, p = 0.0001) and air trapping (ratio of residual volume to total lung capacity [RV/TLC], p = 0.0001). Pa_CO2 significantly decreased (p = 0.003). Pa_O2 tended to increase after LVRS, but this was not statistically significant (p = 0.18).

Table 4 displays pre- and post-LVRS exercise and respiratory muscle function data. At baseline all patients were severely limited in maximal exercise performance as seen by reductions in O₂max, E_max, and exercise time. At peak exercise, patients exhibited a rapid-shallow breathing pattern, with a mean VT of 0.78 ± 0.26 L, and f_b of 31 ± 7 bpm. The baseline f_b/VT ratio at maximal exercise was 43 ± 17. The 6-min walk distance (6 MWD) was 289 ± 91 m.

After LVRS there were significant improvements in E_max (p = 0.0007), exercise time (p = 0.0001), and 6 MWD (p = 0.002), with a trend toward an increase in O₂max (p = 0.08). In addition, at maximal exercise, patients showed a reduction in f_b/VT (34 ± 13; p = 0.005), which was due to an increase in VT (0.97 ± 0.29 L; p = 0.0004) and not decreased f_b (30 ± 6 breaths/min; p = 0.27).

At baseline, patients also showed impairments in global respiratory muscle (PI_max, PE_max), and transdiaphragmatic pressures (Pdi_{max sniff}, Pdi_twitch) (Table 4). Post-LVRS, there were significant increases in PI_max (p = 0.0001), PE_max (p = 0.03), and Pdi_{max sniff} (p = 0.0003), with a trend toward an increase in Pdi_twitch (p = 0.06).

Correlates of Hypercapnia

Resting Pa_CO2 levels pre-LVRS correlated with baseline FEV₁ (p = 0.0001; r = 0.68), RV/TLC (p = 0.02; r = 0.41), and E (p = 0.007; r = 0.47), VT (p = 0.02; r = 0.42), and VD (p = 0.04; r = 0.37) at maximal exercise. No significant relationship was found between Pa_CO2 and age (p = 0.17), percentage of ideal body weight (%IBW) (p = 0.17), or DL_CO/VA (p = 0.09).

Patients who were more hypercapnic pre-LVRS had the greatest reduction in Pa_CO2 after LVRS (Figure 1). In addition, these patients also had the greatest increase in DL_CO/VA (p = 0.009), PI_max (p = 0.03), and a trend toward a correlation with increases in FEV₁ (p = 0.06).

View larger version (13K): [in this window] [in a new window]   Figure 1.   Plot shows that those patients who were more hypercapnic at baseline, had the greatest reduction in PaCO2 after LVRS (r = 0.61, p < 0.001).

To investigate factors that may be responsible for the decrease in Pa_CO2 observed post-LVRS, single and multiple linear regression analyses were performed between the magnitude of change in Pa_CO2 and the percent change in various physiologic and functional parameters listed in Table 5. Figure 2 shows the significant relationship between decreases in Pa_CO2 and increases in airflow, global inspiratory muscle strength, and diffusion capacity. The reduction in Pa_CO2 was also significantly correlated with decreases in hyperinflation post-LVRS. Decreases in Pa_CO2 also correlated with changes in breathing pattern parameters (E_max, VT) during peak exercise as seen in Figures 3A and 3B. Demographic parameters (age, %IBW) showed no correlation with changes in Pa_CO2 post-LVRS.

View larger version (22K): [in this window] [in a new window]   Figure 2.   Correlations between percent change in PaCO2 post-LVRS and percent change in physiologic parameters: (A) FEV1, (B) PImax, (C ) DLCO, and (D) RV/TLC are depicted.

View larger version (13K): [in this window] [in a new window]   Figure 3.   Correlations between percent change in PaCO2 post-LVRS and percent change functional parameters at maximal exercise: (A) Emax and (B) VT.

Multiple linear regression analysis revealed the highest correlation between percent change in Pa_CO2 and the combination of FEV₁, DL_CO/VA, f_b/VT, and PI_max (p = 0.0004, r = 0.76).

We also found significant intersubject variability with respect to changes in Pa_CO2 observed post-LVRS (Figure 4A). Figure 4B shows that the mean percent change in Pa_CO2 was 4%, and the distribution was almost normal.

View larger version (21K): [in this window] [in a new window]   Figure 4.   (A) Individual PaCO2 levels pre- and post-LVRS. (B) Mean percent change PaCO2 post-LVRS.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Following LVRS, our patients demonstrated a mean decrease in Pa_CO2 of 4%. This reduction in Pa_CO2 was strongly correlated with increases in airflow, diffusion capacity, and inspiratory muscle strength, as well as a reduction in the degree of hyperinflation. In addition, decreases in Pa_CO2 post-LVRS also correlated with increases in E and VT during maximum exercise. Patients with the highest baseline resting Pa_CO2 had the greatest percent reduction in Pa_CO2 as well as the greatest increases in DL_CO and PI_max post-LVRS. Moreover, patients who had significant reductions in Pa_CO2 also showed the greatest reduction in RV/TLC post-LVRS.

We found significant correlations between reductions in Pa_CO2 and decreases in hyperinflation as well as increases in respiratory muscle strength. Earlier studies prior to the reintroduction of LVRS that examined the correlation between hypercapnia, lung volume, and respiratory muscle strength corroborate our findings. Rochester and Braun assessed respiratory muscle strength in 32 patients with severe COPD and confirmed the inverse relationship between hyperinflation and global respiratory muscle strength (4). They also found elevated Pa_CO2 in 13 of 18 patients whose PI_max was < 55 cm H₂O. Begin and Grassino examined the relationship between CO₂ retention and lung mechanics in 311 patients with COPD with varying degrees of hypercapnia (5). They found that patients who were more hypercapnic (Pa_CO2 > 55 mm Hg) demonstrated greater degrees of hyperinflation, airflow obstruction, and respiratory pump dysfunction as measured by PI_max. As a result of these mechanical derangements, the hypercapnic group exhibited significantly lower E when compared with normocapnic subjects, mainly due to reductions in VT. Correlations were also found between Pa_CO2 and VD/VT, airflow obstruction, and ratio of pulmonary resistance to maximal inspiratory mouth pressure (RL/PI_max), an index that provides a normalization of mechanical load (total lung resistance) to inspiratory muscle strength. The investigators postulated that alveolar hypoventilation may be a protective strategy employed to avoid respiratory muscle fatigue and failure. This was especially evident in obese patients who have a less compliant respiratory system, and, therefore, more impediments to already compromised inspiratory muscle mechanics.

Similar to the above findings, our patients also exhibited a relationship between baseline resting hypercapnia and VT at maximal exercise, as well as E_max. We were unable to demonstrate any relationship between %IBW and baseline Pa_CO2. The %IBW did not predict the change in Pa_CO2 in our patients post-LVRS, as well. Our patients were well-nourished at baseline, but not overweight, which may have resulted in less mass loading of the chest wall and thus a minimal effect on respiratory pump function and the development of hypercapnia.

Another factor that may cause alveolar hypoventilation is an impaired chemical responsiveness to increasing degrees of hypoxia or hypercapnia. The alteration in chemical responsiveness has been attributed to longstanding hypoxia (26), and/or changes in serum bicarbonate (27). We did not specifically examine this topic in our patients, although most studies agree that a failure of neuromechanical coupling resulting from the aforementioned mechanical derangements is the main factor leading to decreased ventilation. In fact, Gorini and coworkers examined neural drive and respiratory mechanics during room air breathing in normocapnic, mildly hypercapnic (Pa_CO2 < 43 mm Hg), and moderately hypercapnic (Pa_CO2 > 47 mm Hg) patients and demonstrated that neural drive was greater in more severely hypercapnic patients (28). A follow-up study by the same investigators examined CO₂ chemoresponsiveness in hypercapnic and normocapnic patients by the CO₂ rebreathing test and found that mechanical impairment to ventilation and to a lesser extent, inadequate chemoresponsiveness may be responsible for CO₂ retention (29).

Three to 6 mo post-LVRS 28 of our 33 patients demonstrated decreases in Pa_CO2 with a mean decrease of 4%, a value that is concordant with that seen in other studies (7, 15). Other investigators, however, have shown minimal changes in Pa_CO2 post-LVRS (6, 10, 12, 14, 16, 17). There are many reasons for this apparent discrepancy. Many studies excluded patients with Pa_CO2 > 50 mm Hg (16, 18), or even 45 mm Hg (6). Indeed, some investigators have suggested that more severely hypercapnic patients may do worse after surgery (30, 31). Our experience has been quite different. We (32) have previously shown that LVRS can be safely performed even on ventilated patients with severe hypercapnia (mean Pa_CO2 60 mm Hg). We did not exclude patients in this study based upon initial Pa_CO2, and in fact included seven patients whose Pa_CO2 level was > 50 mm Hg. Importantly, we corroborate the findings of others (3, 12) who have shown that it is the subgroup with the highest degree of hypercapnia that demonstrates the greatest post-LVRS reduction in Pa_CO2.

Another factor that may account for the varied responses in Pa_CO2 after LVRS observed by different investigators involves the differences in surgical approach and technique. Unilateral versus bilateral LVRS, as well as various technical factors (laser versus stapling resection) may affect comparisons in conclusions among studies. Also, patient selection may have not been uniform with respect to maximal bronchodilation. Submaximal bronchodilatation may have resulted in elevated levels of physiologic dead space and CO₂ retention. All of our patients underwent bilateral LVRS using stapling resection, and all had evidence of either maximal bronchodilation or irreversible airflow obstruction as evident by a < 5% bronchodilator response.

We are the first to examine the changes in Pa_CO2 and correlate them with improvements in physiologic and functional parameters post-LVRS. It is thought that by removing nonfunctional areas of lung and reducing hyperinflation, LVRS produces the following effects: restoration of elastic recoil of small airways, restoration of normal outward chest wall elastic recoil, improved ventilation-perfusion relationships, and a reduction in end-expiratory lung volume which optimizes respiratory muscle function (9). Significant improvements in respiratory mechanics were achieved after surgery, with increases in FEV₁, Pdi_{max sniff}, PI_max, and reductions in RV/TLC. These improvements are reflected in an increased E at maximum exercise, primarily by an increase in VT, and thus resulting in a reduction in rapid-shallow breathing pattern. As expected, changes in Pa_CO2 were significantly correlated with improvements in mechanics, with positive correlations demonstrated between the change in Pa_CO2 and FEV₁, PI_max, and DL_CO/VA. The change in Pa_CO2 showed an inverse correlation with RV/ TLC. The change in Pa_CO2 also correlated with functional improvements in breathing pattern (less rapid and shallow) and overall ventilation post-LVRS.

This study has important implications for LVRS selection criteria. Hypercapnia has been used to exclude patients from LVRS (6, 7, 13, 15, 16, 18). This practice, however, may be limiting the very subgroup of patients with severe COPD that may derive the most benefit from the surgery. Unfortunately, it is impossible to predict from baseline demographics or physiologic status which patients will show the most marked reduction in Pa_CO2 postsurgery, but it is clear that Pa_CO2 alone is not predictive.

The overall clinical significance of a mean 4% decrease in Pa_CO2 post-LVRS is unclear. In the subset of 11 patients with baseline Pa_CO2 > 45 mm HG, a decrease in Pa_CO2 post-LVRS may have resulted in an increase in oxygenation. In patients eucapnic at baseline, a reduction in Pa_CO2 post-LVRS would have little to negligible improvement in oxygenation, and thereby only signify an improvement in ventilatory mechanics. Additional study is required to determine the independent effect of post-LVRS reduction in Pa_CO2 on clinical status.

Why do some patients decrease Pa_CO2 and others do not? We postulate that patients who at baseline have similar degrees of hyperinflation, airflow obstruction, and respiratory muscle strength impairment and who are able to increase ventilation as a result of LVRS benefit the most. That increased alveolar ventilation may be pivotal in reducing CO₂ retention has been alluded to by other researchers as well (10). We have demonstrated that functional improvements in lung mechanics play a major role in the reduction in Pa_CO2 observed post-LVRS, by allowing the respiratory system to operate more efficiently as a pump to eliminate CO₂.

We did not specifically address how changes in neural drive that occur postsurgery may affect alveolar ventilation and Pa_CO2; however, two recent studies (6, 33) have demonstrated reductions in measurements of central drive. Neither study, however, found correlations between these measurements and Pa_CO2, suggesting again that neuromechanical coupling, specifically, derangements in respiratory pump function, are the most important factors leading to CO₂ retention.

In summary, we found that reductions in Pa_CO2 after LVRS in patients with severe COPD correlated with improvements in respiratory mechanics leading to greater E. Baseline Pa_CO2 is not predictive of functional or physiologic outcome, and should not, therefore, be used alone to exclude patients from LVRS.