INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Increased airflow resistance, hyperinflation, and air trapping are the main pathophysiologic features of chronic obstructive pulmonary disease (COPD) (1). Hyperinflation correlates directly with an increased resting pleural pressure, and is associated with decreased diaphragm activation and increased expiratory muscle recruitment (2). Hyperinflation also affects respiratory mechanics by passively shortening the operating length of the diaphragm and changing its linkage with the rib cage, thus placing the diaphragm at mechanical disadvantage (3). This affects respiratory muscle strength (4, 5) and volume displacement, thus contributing to an impairment of exercise tolerance. The mass, thickness, and area of the diaphragm are also altered in COPD (6), as a result of general skeletal muscle wasting commonly seen in these patients. These morphometric changes in the diaphragm, and especially in its operating length, have been postulated to contribute to the reduced inspiratory muscle strength observed in COPD patients (5, 12, 13).

In select COPD patients, lung volume reduction surgery (LVRS) improves spirometric results and lung volumes (14), alters the breathing pattern during exercise (17), and increases diaphragm strength (20), leading to reduction of dyspnea and improvement in exercise performance. Although it has been hypothesized that LVRS leads to an increase in diaphragm strength by reducing end-expiratory lung volume, and thereby increases diaphragm precontraction length, none of the published studies, including our prospective controlled trial (22), has been able to demonstrate a significant correlation of increased diaphragm length with a reduction in lung volume. We therefore conducted an investigation to determine: (1) the direct effect of LVRS on diaphragm length and the area of apposition of the diaphragm with the chest wall in patients with severe COPD; (2) whether alterations in diaphragm length correlate with changes in diaphragm strength after LVRS; and (3) the relationship of an increase in diaphragm length after LVRS with exercise performance.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patient Selection

Our study was done with 25 consecutive patients who underwent LVRS at Temple University Hospital. All patients enrolled in our LVRS program had severe emphysema, were receiving optimal medical therapy (bronchodilators, inhaled and/or oral corticosteroids, home oxygen therapy, and comprehensive outpatient pulmonary rehabilitation), and were clinically stable, with no evidence of acute respiratory symptoms. The patients had no evidence of kyphoscoliosis or history of chest trauma; did not show any thoracic or parenchymal abnormalities on routine chest roentgenography, other than those consistent with a diagnosis of COPD; and fulfilled the inclusion criteria shown in Table 1. All patients signed informed consents forms, as approved by the Temple University Institutional Review Board for Human Research.

All patients had plain chest roentgenograms; underwent a variety of pulmonary function studies (e.g., spirometry, lung volumes measured by plethysmography, diffusion capacity); had an incremental, symptom-limited maximum exercise test; and had measurements of diaphragm strength made within 3 wk before LVRS and 3 mo after LVRS. For various reasons (exacerbations, inability to tolerate a test, illness unrelated to COPD, lower extremity fracture), some of the patients were unable to complete all the physiologic testing at exactly the same time as roentgenography was done. Seven of the patients had plain chest roentgenograms made within a year prior to their baseline presurgical evaluation, and these served as reference data.

Normal Controls

In addition to the studies described previously, we made diaphragm length measurements in eight normal controls who had undergone routine preoperative chest radiography, in order to compare our technique of diaphragm length measurement to that reported by Braun and coworkers (13).

Physiologic Measurements

Pulmonary function testing. Pulmonary function testing was performed with a System 6200 Autobox DL Plethysmograph (SensorMedics, Yorba Linda, CA), using American Thoracic Society guidelines (23). FVC, FEV₁, and FEV₁/FVC were measured. Thoracic gas volumes were measured in a body plethysmograph. All postbronchodilator pulmonary function data are reported in absolute numbers and as percents of normal predicted values (23).

Diaphragm strength. Transdiaphragmatic pressure (Pdi) was measured as previously described (24, 25) with two balloon-tipped catheters placed into the distal esophagus (endoesophageal pressure [Pes]) and stomach (gastric pressure [Pga]), and connected to pressure transducers (100 ± 5 cm H₂O; Validyne, Ventura, CA). The pressure waveforms were continuously displayed on a strip-chart recorder (ES 1000; Gould, Dayton, OH). Pdi was calculated as the difference between end-inspiratory and end-expiratory values.

Maximal Pdi was determined with volitional and nonvolitional techniques. The volitional technique consisted of from three to six maximal sniff maneuvers (Pdi_{max sniff}). After each patient could reproducibly achieve several maximal efforts, a total of three values, all within 5% of one another, were averaged and reported. Nonvolitional measurements were made through bilateral phrenic nerve twitch stimulation (Grass S88 Stimulator, Quincy, MA; 100 to 140 V, 0.1 ms duration). From six to fifteen consecutive twitches, separated by pauses of at least a 3-s, were delivered to each patient, with three to five readings, all within 5% of each other, averaged and reported as Pdi_twitch. All measurements were made from FRC by monitoring the Pes waveform with patients seated in the upright position.

Exercise testing. Patients underwent incremental, maximal treadmill exercise (Precor, 9.4 sp; Precor Inc., Bothell, WA) starting at a 0% incline and 1 mph. The incline was increased by 3% and the speed by 1.5 mph every 3 min until symptom-limited maximum exertion was reached. Oxygen uptake (O₂), carbon dioxide production (CO₂), minute ventilation (E), tidal volume (VT), and respiratory rate (RR) were recorded with a metabolic cart (SensorMedics 2900). Transcutaneous oxygen saturation (Nellcor N-200, Chula Vista, CA) and a multiple-lead electrocardiogram (ECG) (ECG Horizon; SensorMedics) were continuously recorded.

Roentgenographic Measurements of Diaphragm Length

Diaphragm length was measured with plain posteroanterior (PA) and left lateral chest roentgenograms. These were made within 3 wk before LVRS and 3 to 6 mo after LVRS. As noted earlier, seven patients had chest roentgenograms made previously, within a year of their preoperative evaluation, which served as reference data.

Each PA and lateral chest roentgenogram was produced with a General Electric MPX-80 roentgenography unit (Milwaukee, WI) with a 1,000 mA generator. The tube focal point was 6 ft from the roentgenograph cassette. This distance was consistent for every roentgenogram obtained during this investigation, in order to ensure that the magnification of structures did not influence the diaphragm length measurements. Initial settings were 140 kUP, 500 mA, and an exposure time of 0.32 s. The X-ray beam was centered in a standard fashion: on the midsternum at the xiphoid level in the PA view, and on the midhemithorax at the xiphoid level in the lateral view. Two techniques were used to assess diaphragm length, as outlined subsequently.

Diaphragm Length

To assess the length of the diaphragm (total diaphragm length and the length of its zone of apposition with the chest wall), we used the technique described by Braun and coworkers (13). Roentgenograms were made at maximal inspiration, with patients standing upright with their arms at their sides, and firmly positioned against a backboard.

On the PA view (Figure 1), the length of the right hemidiaphragm silhouette (PADL) was measured from its lateral intersection with the chest wall (according to the lateral rib or interspace) to its intersection with the vertebral column. On the lateral view the length of the right hemidiaphragm silhouette (LDL) was measured from the spine to the sternum. Total diaphragm length (TDL) refers to the sum of PADL and LDL. To assess the area of apposition of the diaphragm with the chest wall, the length of the vertically oriented portion of the right hemidiaphragm muscle (VDML) was also measured, from the lateral intersection with the chest wall to the point where the diaphragm silhouette became horizontal.

View larger version (11K): [in this window] [in a new window]   Figure 1.   Diaphragm length measurements on PA and lateral chest roentgenograms. (A to C ) Right diaphragm length on PA view (PADL). (A to B) Diaphragm area of apposition with chest wall (VDML). (D to E ) Right diaphragm length on lateral view (LDL).

All measurements were made with a nonelastic string to follow the contour of the diaphragm, and were recorded in centimeters. To standardize comparisons among patients, all the diaphragm lengths described were divided by height and expressed as the corresponding diaphragm length indices (cm/cm).

Diaphragm Height

In addition to measuring diaphragm length from the sum of PADL and LDL, we also indirectly measured diaphragm length by determining the height of its dome in the thorax before and after LVRS. In essence, a more cephalad position of the diaphragm dome was indicative of diaphragm lengthening. Roentgenograms were made at maximal inspiration, with patients standing upright with their arms at their sides, and positioned firmly against a backboard (26).

To standardize for artifact that can occur with changes in roentgenographic technique, all measurements were referenced to the bony thorax by two methods (Figure 2). On the PA view: (1) diaphragm dome height was assessed from the distance between horizontal lines drawn at the level of the manubrium sterni and the highest point of the diaphragmatic dome, measured at the midclavicular line (cm); and (2) the highest point of the diaphragmatic dome was referenced horizontally to the vertebral column. The superior aspect of the first thoracic vertebral body was assigned a value of zero. Each subsequent vertebral body was assigned a value of 0.8 units, and the subjacent interspace was assigned a value of 0.2 units (26).

View larger version (13K): [in this window] [in a new window]   Figure 2.   Diaphragm dome height measurements on PA and lateral chest roentgenograms. (A to B) and (D to E ) Distance between right and left hemidiaphragm domes and a horizontal line at the level of the manubrium sterni, respectively (on PA view). (G to H ) Distance between right and left hemidiaphragm domes and a horizontal line at the level of the manubrium sterni (on lateral view). (A to C ) and (D to F ) Horizontal references to the vertebral column (on PA view). (G to I ) Horizontal references to the vertebral column (on lateral view).

The same two measurement methods were also used to assess diaphragm dome height on the lateral view.

Surgical Technique

LVRS was performed via median sternotomy and bilateral stapling resection. The goal of resection was to remove 20 to 40% of the volume of one lobe from each lung. High-resolution chest computed tomography (CT) and quantitative ventilation/perfusion scans were used preoperatively to target lung regions with the worst emphysema (i.e., areas of greatest gas trapping with poorest perfusion). At the end of the operation, chest tubes were placed and managed in the conventional manner.

Data Analysis

All measurements were made in centimeters and rounded off to the nearest tenth of a centimeter. Every measurement was made independently by two investigators, and correlated within a tenth of a centimeter. Measurements made on preoperative and postoperative roentgenograms and measurements of physiologic parameters were made on separate days. At the time of the postoperative measurements, the investigators were blinded to the results of both the preoperative roentgenographic and pre- and postoperative physiologic measurements.

All data are expressed as mean ± SD, except where otherwise noted. Statistical analysis was done with the paired Student's t test in comparing preoperative and postoperative values. Linear regression analyses were used to evaluate correlations between roentgenographic and physiologic measurements. Values of p < 0.05 were considered statistically significant. All statistical analyses were done with a commercially available computer software program (Sigmastat, version 1.0; Jandel Corp., San Rafael, CA).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patient Characteristics

From November 1994 to October 1996, 69 patients with severe COPD underwent LVRS. Twenty five patients had pre- and postoperative PA and lateral chest roentgenograms. These patients constituted the study cohort.

Baseline characteristics of all patients in the study cohort are shown in Table 2. Patients were 58 ± 8 yr of age; 11 were male and 14 were female. As a group, the patients in the cohort were not undernourished (ideal body weight: 109 ± 11% predicted). All patients were functionally limited, being in New York Heart Association (NYHA) Class III or IV. The majority were oxygen dependent, had a significant smoking history, were hyperinflated with severe airflow obstruction, and had abnormal gas exchange. All patients finished 8 wk of outpatient pulmonary rehabilitation before LVRS. All preoperative baseline data were recorded after completion of rehabilitation.

Physiologic Data

Results of lung function studies, exercise testing, and diaphragm strength measurements (spirometry, lung volumes, 6MWD, maximum E (E_max), maximum O₂ (O₂max), Pdi_{max sniff}, and Pdi_twitch) at baseline and 3 to 6 mo after LVRS are shown in Table 3. All patients had severe airflow obstruction, moderate to severe air trapping and hyperinflation, and decreased exercise capacity.

At 3 to 6 mo after LVRS, FEV₁ improved to 0.87 ± 0.29 L (p < 0.001), TLC decreased to 6.6 ± 1.4 L (p < 0.001), RV decreased to 3.6 ± 1.1 L (p < 0.0001), E_max increased to 27.0 ± 6.2 L/min (p < 0.003), O₂max increased to 14.9 ± 3.6 ml/kg/min (p < 0.001), and 6MWD increased to 330 ± 84 m (p < 0.001). There were also significant improvements in mean Pdi_{max sniff} and Pdi_twitch.

At 3 to 6 mo after LVRS, the patient cohort showed no significant change in nutritional status (ideal body weight: 112 ± 9% predicted).

Roentgenographic Data

Diaphragm length. The following measurements were recorded in the eight normal controls (three male and five females), aged 35 ± 11 yr (range: 18 to 49 yr) PADL: 14.6 ± 1.2 cm (range: 12.4 to 16.2 cm) LDL: 21.5 ± 3.0 cm (range: 17.5 to 25.9 cm); and VDML: 5.2 ± 0.8 cm (range: 4.4 to 6.9 cm).

Right hemidiaphragm lengths before and after LVRS in the COPD patients are summarized in Table 4. As compared with baseline, there was a significant increase in right hemidiaphragm length after LVRS as measured on the PA view (from 13.9 ± 1.9 cm to 14.5 ± 1.7 cm; p = 0.02). When the vertical, muscular portion of the diaphragm was measured specifically, there was an even more significant increase in its length after LVRS (from 2.08 ± 1.5 cm to 3.00 ± 1.6 cm; p = 0.01). These changes remained significant when corrected for height. However, the right hemidiaphragm length as measured on the lateral view, and the total length represented by the sum of the two views, did not show significant changes after LVRS (from 20.5 ± 3.0 cm to 19.9 ± 3.6 cm; p = 0.1; and from 34.6 ± 4.1 cm to 34.8 ± 4.5 cm; p = 0.6, respectively).

Diaphragm height. The heights of both hemidiaphragm domes before and after LVRS are given in Table 5. Both techniques of diaphragm height measurement showed a highly significant cephalad displacement of the domes of each hemidiaphragm after LVRS as compared with baseline. Whether measured as the distance between the diaphragm domes and the manubrium sterni, or as the number of thoracic vertebral-body units, the increase in dome height confirmed an increase in total diaphragm length after LVRS.

Reference Data

Seven of the 25 COPD patients had plain chest roentgenograms done within a year before their baseline presurgical evaluation, which served as reference data. No significant changes in diaphragm length or area of apposition were seen as measured with either the nonelastic string technique (Table 6) or the diaphragm dome height technique (Table 7) when preoperative measurements were compared with reference films. This contrasts with the significant lengthening of the diaphragm seen from 3 to 6 mo after LVRS.

Correlations

To evaluate the clinical significance of our findings, we correlated the described change in diaphragm length after LVRS with physiologic parameters. Figures 3 and 4 demonstrate strong correlations between change in lung volume and change in diaphragm length after LVRS. Figures 3A and 3B, showing results with the diaphragm dome height measurement technique, suggest a direct relationship between the reduction in TLC and increase in diaphragm length (R = 0.47, p = 0.04 for the right hemidiaphragm in the PA view; R = 0.70, p = 0.006 for the right hemidiaphragm in the lateral view; R = 0.55, p = 0.02 for the left hemidiaphragm in the PA view; and R = 0.72, p = 0.003 for the left hemidiaphragm in the lateral view). Similarly, the direct technique of measuring diaphragm length showed a significant correlation between the reduction in TLC and increase in diaphragm length (R = 0.55, p = 0.04) (Figure 4A), and a trend toward a significant correlation between the reduction in RV and increase in diaphragm length (R = 0.50, p = 0.07) (Figure 4B).

View larger version (24K): [in this window] [in a new window]   View larger version (23K): [in this window] [in a new window]   Figure 3.   Correlations between percent change in TLC after LVRS and percent changes in right (A) and left (B) hemidiaphragm dome heights (on PA and lateral chest roentgenograms). Diaphragm dome height is defined as the distance between the dome and the horizontal line at the level of the manubrium sterni (i.e., a negative change indicates the diaphragm moving more cephalad).

View larger version (17K): [in this window] [in a new window]   Figure 4.   Correlations between percent change in diaphragm length (PADL) after LVRS and percent change in TLC (A), and RV (B).

To evaluate the potential significance of an increased diaphragm length on diaphragm strength and exercise capacity, correlations with changes in diaphragm strength (Pdi_{max sniff}, Pdi_twitch), exercise parameters (O₂max, E_max), and maximum voluntary ventilation (MVV) were made. Figure 5 demonstrates a strong relationship between the increase in diaphragm length and the increase in Pdi_{max sniff} (R = 0.79, p = 0.01) after LVRS. A trend toward a significant correlation with Pdi_twitch (R = 0.69, p = 0.09) is also seen. Figure 6 shows significant correlations between the diaphragm lengthening after LVRS and increases in O₂max and E_max (R = 0.70, p = 0.01, and R = 0.71, p = 0.01, respectively). Figure 7 shows a strong positive correlation between the increase in diaphragm length and increase in MVV after LVRS (R = 0.73, p = 0.0008).

View larger version (18K): [in this window] [in a new window]   Figure 5.   Correlations between percent change in diaphragm length (PADL) after LVRS and percent change in diaphragm strenth (Pdimax sniff, and Pditwitch).

View larger version (17K): [in this window] [in a new window]   Figure 6.   Correlations between percent change in diaphragm length (PADL) after LVRS and percent change in exercise parameters ( O2max and Emax).

View larger version (21K): [in this window] [in a new window]   Figure 7.   Correlations between percent change in diaphragm length (PADL) after LVRS and percent change in MVV.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Our data confirm findings of prior investigations that LVRS improves spirometric parameters, lung volumes, and diaphragm strength in select patients with severe COPD (14, 22, 27). They also clearly demonstrate an increase after LVRS in diaphragm length as measured with three different techniques. The increase in diaphragm length was most pronounced in the area of apposition of the diaphragm with the rib cage. This diaphragm lengthening seems to be directly related to the reduction in lung volume seen after LVRS. Our data also show the significant relationship that increases in diaphragm length have with improvement in maximum exercise performance and MVV after LVRS.

Several uncontrolled trials, with small numbers of patients, have found improvements in diaphragm strength following LVRS (18, 20, 28). Teschler, Martinez, and Tschernko and their colleagues have demonstrated increases in transdiaphragmatic pressure at 1, 3, and 6 mo after LVRS, respectively (18, 20, 21). Martinez and coworkers also documented significantly lower resting pleural and gastric pressures after LVRS, which correlated with improvement in a dyspnea index (18). Benditt and associates found similar decreases after LVRS in pleural and gastric pressures during restful breathing, and improved inspiratory pressure generation by the diaphragm during exercise (17).

Although the foregoing studies showed that diaphragm strength improved after LVRS, they were unable to elucidate a specific mechanism for this improvement, and failed to demonstrate a significant correlation with a reduction in lung volume. We recently conducted a large, prospective, controlled trial, using a variety of volitional and nonvolitional (i.e., electrophrenic stimulation) techniques to measure transdiaphragmatic pressures in order to evaluate changes in diaphragm strength before and after LVRS (22). Although there was no change in diaphragm strength after 8 wk of pulmonary rehabilitation, we found a significant improvement in diaphragm strength at 3 mo after LVRS. However, as in prior investigations, we also did not find a relationship between increases in transdiaphragmatic pressure generation and a reduction in lung volume.

Failure to correlate an improvement in diaphragm strength with LVRS in the studies just described may have been due to several factors. Although several studies have demonstrated a linear inverse correlation of diaphragm strength with lung volume over the range of VC in normal individuals (29), scant evidence exists for the linearity of this relationship in patients with severe COPD and hyperinflation (3, 31). In patients with severe hyperinflation, in whom measured diaphragm pressures are lowest, variability or errors in the measurement of lung volumes or transdiaphragmatic pressures may significantly impair the ability to determine the effect of small to moderate reductions in lung volume on small to moderate increases in diaphragm strength.

Studies examining transdiaphragmatic pressures before and after LVRS have either used maximum sniff or nonspecified techniques of assessment (17, 20), which may be responsible for the variability in diaphragm strength measurements and failure to correlate diaphragm strength with changes in lung volume after LVRS. However, even when we used the combined expulsive-Mueller maneuver with oscilloscopic visual feedback (34) to ensure maximal and reproducible volitional measures of maximum transdiaphragmatic pressures, or bilateral supramaximal electrophrenic twitch stimulation for nonvolitional assessment, we failed to correlate a reduction in lung volume with the increases in diaphragm strength after LVRS. Laghi and colleagues also failed to show a correlation of increases in Pdi_twitch after LVRS and reductions in lung volume (28). These observations tend to suggest that variability or limitations in the technique of measuring diaphragm strength among studies does not appear to be the sole factor responsible for the failure to correlate reductions in lung volume with increases in diaphragm strength after LVRS.

Although lung volume measurement by body plethysmography is a reproducible technique in normal individuals (35), this may not be true in patients with severe airflow obstruction. Several studies have demonstrated the variability of lung volume determinations, even with body plethysmography, in patients with stable airflow obstruction (36, 37). Significant variability in lung volume determination as a result of fluctuations in airflow obstruction from airways disease, or from a prolonged expiratory time constant negating mouth pressure as representative of alveolar pressure, may seriously undercut the ability to correlate increases in diaphragm strength with changes in lung volume. Because a reduction in lung volume is believed to mediate an increase in the generation of Pdi through an increase in diaphragm length, perhaps a more direct approach that would obviate the variability in lung volume determination in this patient group would be to directly measure changes in diaphragm length, as was done in the present study.

In this study we showed directly that LVRS increases diaphragm length, especially in that portion of the muscle directly apposed to the inner portion of the lower rib cage, the zone of apposition. As suggested by Mead and Loring, contraction of the diaphragm muscle fibers in the zone of apposition may be responsible for the majority of diaphragmatic volume displacement during respiration (38). The size of the zone of apposition has been shown to correlate with the magnitude of force transmitted to the lower rib cage and the increase in abdominal pressure with diaphragmatic contraction (39). Therefore, an increase in the zone of apposition for a given increase in abdominal pressure would result in a greater expanding effect on the transverse lower rib cage during diaphragmatic contraction.

In contrast to the significant diaphragm lengthening we found with measurements made in the transverse plane, we did not find any change in diaphragm length as measured in the anteroposterior (AP) dimension (Table 4). There are two possible explanations for this. First, the diaphragm is dome shaped, and its lengthening after LVRS may occur in any one of the radii emanating from its tendinous central portion, which can be underestimated with two-dimensional measurements. Second, a change in the configuration of the bony thorax after LVRS may influence the measurement of diaphragm length. We have recently shown that LVRS alters the configuration of the bony thorax in COPD patients, by demonstrating a significant reduction in the lower AP rib cage diameter but not lower in the transverse diameter at 3 and 12 mo after LVRS (40). A reduction in the AP dimension of the bony rib cage should move the posterior and anterior diaphragm insertion points inward and elevate the dome of the diaphragm in a more cephalad direction. These complex structural changes may not have resulted in an appreciable lengthening of the diaphragm as measured in the lateral view with our roentgenographic techniques.

Our study has several limitations. First, measuring diaphragm length with plain chest roentgenography is subject to error. On the PA view it can sometimes be difficult to locate the median segment of the right hemidiaphragm contour. On the lateral view, the same difficulty exists for the vertebral portion of the contour. However, this tends to be more of a problem in films made at lower lung volumes (FRC or RV) than in those made at TLC. All the roentgenograms in the present study were made at maximal inspiration, and all measurements were reproduced by two blinded, independent investigator reviews before and after LVRS.

Another potential limitation of our study was the lack of objective monitoring to ensure maximal inspiration at the time roentgenograms were made. However, an effort was made to instruct and encourage the patients to maintain a maximal inspiration. Additionally, the erect position of patients at the time of roentgenographic exposure may have differed from the "normal" posture for some of the patients, and may have contributed to the altered relationship of the rib cage to the diaphragm. This should not have had a significant influence on our results, since all comparisons were made in the same individuals studied repeatedly, rather than in different groups of individuals examined at separate study points.

The data for our normal controls vary somewhat from those of Braun and colleagues (13), possibly as a result of differences in radiographic technique, or image magnification, or slight variation in diaphragm length measurements due to our interpretation of their method. However, the techniques we used for chest imaging and diaphragm length measurement before LVRS were identical to those after LVRS, and should not have significantly influenced our findings.

There are several strengths to our study design. The study was done prospectively, with two blinded, independent investigators making measurements with three different methods. The nonelastic string method was chosen because it is a simple, noninvasive way of measuring diaphragm length with a plain chest roentgenogram, and has been shown to correlate well with necropsy measurements (13). The technique of measuring the change in diaphragm dome height was chosen to provide a fixed reference for diaphragm height that was normalized to the bony thorax. The seven patients providing reference data helped to solidify our results by showing that the increase in diaphragm length found after LVRS was not merely due to variations in two roentgenograms taken at different time points.

We believe that the present study adds useful information to the literature by convincingly showing an increase in diaphragm length after LVRS. However, it does not provide a mechanism for how the diaphragm increases its length. Animal studies may be helpful in postulating a mechanism. Farkas and Roussos (41) and Kelsen and coworkers (42), using emphysematous hamsters, demonstrated that the decrease in diaphragm length seen in this disease state was mainly due to a reduction in the number of sarcomeres and a decrease in sarcomere length. Whether lengthening of the diaphragm after LVRS represents lengthening of the preexisting sarcomeres, remodeling of the diaphragm, or a change in the projected geometry of the diaphragm is unknown. Any or all of these mechanisms are possible and can explain the diaphragm lengthening effect; however, only future studies using diaphragm histology can address the specific mechanism(s) responsible for diaphragm lengthening.

In conclusion, we used roentgenography to evaluate changes in diaphragm length in COPD patients undergoing bilateral LVRS. Using three methods of measurement, we showed a significant increase in diaphragm length and in the area of apposition of the diaphragm with the rib cage after LVRS. Although the mean increase in diaphragm length was small, it is consistent by all three measuring methods, and might have been underestimated with two-dimensional measurements. By showing a direct relationship with physiologic measurements, we provide a potential mechanism for improvement in respiratory mechanics after LVRS, and suggest that: (1) the diaphragm lengthens after LVRS; (2) the diaphragm lengthening found after LVRS is related to a reduction in lung volume; and (3) after LVRS the diaphragm operates at a more optimal position along its length-tension curve, thereby increasing its force-generating capacity and leading to an improvement in exercise capacity and tolerance.