INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

In the initial article describing bilateral lung volume reduction surgery (LVRS) done by median sternotomy for patients with emphysema, Cooper and colleagues (1) reported that the mean ± SD Pa_O2 increased from 64 ± 6.5 mm Hg (n = 18, breathing air) to 70 mm Hg (n = 8, no SD reported, p < 0.05) with no change in the Pa_CO2 when both were remeasured 6 mo following the operation. Little and colleagues (2) observed similar but less marked changes in 14 patients who either had giant bullectomy (n = 3), LVRS via sternotomy with staple resection (n = 6) or unilateral thoracoscopic laser ablation (n = 5) (room air Pa_O2 increasing from 59 ± 2 to 63 ± 2, p < 0.05, no change in Pa_CO2). Similar trends were reported by Daniel and colleagues (3) in 17 patients, although the changes in Pa_O2 and Pa_CO2 did not reach statistical significance. The lack of change in Pa_CO2 in these studies implies that LVRS improves oxygenation without improving alveolar ventilation, suggesting that ventilation-perfusion heterogeneity is reduced.

The results of other studies, however, would link much of the change in oxygenation following LVRS to an increase in alveolar ventilation. Miller and colleagues (4) reported that the mean Pa_O2 increased from 62 to 70 mm Hg (n = 40), while the mean Pa_CO2 decreased from 43 to 40 mm Hg. Similar findings were reported by Tschernko and colleagues (5) (Pa_O2 increasing from 56 ± 3 to 66 ± 3 mm Hg with Pa_CO2 decreasing from 45 ± 1 to 38 ± 1 mm Hg, n = 12).

There are similar disparities in the effect of unilateral LVRS on gas exchange. Roué and colleagues (6) found that the improvement in Pa_O2 was largely the result of an improvement in alveolar ventilation (i.e., Pa_O2 increasing from 54 ± 7 to 61 ± 8 mm Hg with Pa_CO2 decreasing from 45 ± 5 to 41 ± 7 mm Hg, n = 13), but Naunheim and colleagues (7) reported that Pa_O2 increased from 59 ± 9 to 67 ± 11 mm Hg while Pa_CO2 did not change (n = 12). Both Keenan and colleagues (8) and Sciurba and colleagues (9) reported that Pa_O2 did not change while Pa_CO2 decreased (n = 40 and 20, respectively). Hazelrigg and colleagues (10) found no significant changes in either the Pa_O2 or the Pa_CO2 (n = 91, with results from some of the patients included in other reports cited).

The purpose of this study was to measure the effects of LVRS (done through a median sternotomy with stapled resection) on arterial blood gases and to relate the changes observed to a number of theoretical outcomes from the operation.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Between October 1994 and June 1996, 76 patients with severe airflow limitation, airtrapping, and roentgenographic evidence of emphysema underwent LVRS at our institution. Screening criteria for entry into our program included: (1) evidence of severe airflow limitation (15% < FEV₁ < 35% predicted); (2) evidence of severe hyperinflation (total lung capacity > 120% or residual volume > 150% predicted); (3) age < 75 yr; (4) smoking cessation for at least 3 mo; (5) absence of substantive daily sputum production; (6) computed tomographic evidence of emphysema; and (7) absence of other comorbid conditions. Any patient with evidence of pulmonary hypertension by electrocardiogram, chest roentgenograms, or physical examination underwent echocardiography to evaluate pulmonary arterial pressures; if the echocardiogram revealed pulmonary hypertension or was equivocal, right heart catheterization was performed. No patient underwent pulmonary rehabilitation either before or after the operation.

Arterial blood gases were measured with the patients slightly recumbent (i.e., approximately 75°), breathing air for a minimum of 10 min before and 3 mo after bilateral LVRS, which was performed through a median sternotomy with buttressed stapling as described by Cooper and colleagues (1). The alveolar-to-arterial oxygen tension difference (AaPO₂) was calculated using the measured Pa_O2 with the alveolar oxygen tension (PA_O2 calculated from the alveolar gas equation assuming r = 0.8: PA_O2 = PI_O2  Pa_CO2/R, where R = gas exchange ratio and PI_O2 = partial pressure of inspired oxygen.

Data are presented as means ± SD, comparisons are made by paired t testing, and correlations were sought using linear regressions (p < 0.05 was considered significant).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The demographics and pulmonary function tests of the 46 patients studied are presented in Table 1. The degree of airflow limitation and gas trapping observed in our population was similar to that of patients undergoing LVRS in other institutions (1). The population represents 61% of a consecutive series of 76 patients undergoing LVRS at the University of Washington Medical Center between September 1994 and June 1996. Arterial blood gases were measured at mean ± SD 113 ± 88 d (i.e., a mean of approximately 3 mo prior to the operation).

The changes in Pa_O2 and Pa_CO2 measured 3 mo following LVRS are presented in Figure 1. The mean change in Pa_O2 was +3 ± 10 mm Hg (p = 0.58), but the range of change was considerable (17 to +29 mm Hg). The mean ± SD change in Pa_CO2 was 3 ± 5 mm Hg (p < 0.05) with a range of 11 to +5 mm Hg. The mean ± SD change in AaPO₂ was +1 ± 11 mm Hg (p = NS) with a range of 24 to +23 mm Hg. There was no correlation between the change in Pa_CO2 and the change in Pa_O2 or AaPO₂ (Figures 1 and 2).

View larger version (14K): [in this window] [in a new window]   Figure 1.   Changes in PaCO2 and PaO2 3 mo following lung volume reduction surgery (all measurements obtained while patients were breathing air).

View larger version (15K): [in this window] [in a new window]   Figure 2.   Changes in AaPO2 and PaCO2 3 mo following lung volume reduction surgery (all measurements obtained while patients were breathing air).

We found no significant correlation between the changes in Pa_CO2, Pa_O2, or the AaPO₂ and the preoperative Pa_CO2, Pa_O2, AaPO₂, FEV₁, residual volume (RV), total lung capacity (TLC) (all expressed in liters or as percent predicted), or the diffusing capacity for carbon monoxide (DL_CO). Changes in pulmonary function resulting from LVRS are summarized in Table 2. We found no correlation between the changes in Pa_CO2, Pa_O2, or AaPO₂ and the changes in FEV₁, RV, or TLC (measured in liters or in percent predicted), or with the change in the DL_CO occurring as a result of the operation.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The important findings of this study are that (1) LVRS can improve or worsen arterial blood gases; (2) although the mean changes in Pa_O2 and Pa_CO2 are minimal, the range of change in blood gases is considerable; (3) on average, there seems to be a small improvement in alveolar ventilation; and (4) the major beneficial and detrimental effects of LVRS on Pa_O2 seem to occur via changes in ventilation-perfusion heterogeneity.

We were able to obtain follow-up studies on only 61% of patients who were eligible at the 3-mo time point despite contacting those who elected not to return on several occasions. Although more complete and/or more prolonged follow-up might alter the mean changes in the Pa_CO2, Pa_O2, and/or the AaPO₂ observed, it will not alter the observation that these variables can both improve and worsen as a result of LVRS, and that the range of change is considerable. Changes in gas exchange seen at 3 mo may under- or overestimate what occurs later. Although the effect of LVRS on lung mechanics seems to peak at 6 mo (11), the effect on gas exchange may reach a maximum at 3 mo (4).

We assumed that R was equal to 0.8 when calculating the AaPO₂ both before and after LVRS. In an earlier study on the effects of LVRS on cardiopulmonary exercise physiology, we measured resting R values of 0.81 ± 0.79 and 0.83 ± 0.78 in 24 patients before and after the operation (12). Accordingly, our assumption seems justified.

Gas Exchange Abnormalities in Patients with Chronic Obstructive Pulmonary Disease

At a constant barometric pressure and fraction of inhaled O₂, changes in the Pa_O2 can result from alterations in shunt, O₂ uptake, diffusion capacity, alveolar ventilation (A), ventilation-perfusion heterogeneity, and/or mixed venous O₂ tension (P_O2).

Alterations in shunt and/or diffusion. Wagner and colleagues (13) used the multiple inert gas elimination method to characterize gas exchange in 23 patients with chronic obstructive pulmonary disease (COPD). Three patterns of ventilation- perfusion heterogeneity were observed: (1) extensive regions of high ventilation-perfusion (i.e., ventilation-perfusion > 3), with few regions of low ventilation-perfusion and no shunt (i.e., the high "H" pattern); (2) large regions of low ventilation-perfusion with few regions of high ventilation-perfusion and no shunt (i.e., the low or "L" pattern); and (3) a pattern consisting of both low and high ventilation-perfusion regions, again with no shunt. Diffusion abnormalities were not found to contribute to the hypoxemia. Almost all of their patients with emphysema (as opposed to those with bronchitis) had the H pattern of ventilation-perfusion heterogeneity. Accordingly, of the pathophysiologic mechanisms by which LVRS might improve the Pa_O2, reductions in shunt and/or an improvement in diffusion can be eliminated as patients with emphysema have little-to-no shunt, and abnormal diffusion does not contribute to the hypoxemia observed, at least at rest (13). The Pa_O2 could decrease, however, if shunt developed or if diffusion worsened.

Change in alveolar ventilation. The change in Pa_CO2 of 3 ± 5 mm Hg that we observed indicates that, on average, patients increased their alveolar ventilation and/or decreased their CO₂ production following LVRS. We have previously observed (12, 14) that resting minute ventilation and CO₂ production were unchanged following LVRS, but that dead space volume/tidal volume (VD/VT) decreased slightly but significantly (4%, p < 0.05). Accordingly, the decrease in Pa_CO2 that we observed can be attributed to a reduction in VD/VT that results from LVRS.

Changes in ventilation-perfusion heterogeneity. Although significant mean changes in Pa_O2 and AaPO₂ were not observed, many patients had considerable changes in these variables, indicating that LVRS can worsen or improve ventilation/perfusion heterogeneity. The mechanisms by which these opposing effects might occur can be demonstrated by using the CO₂-O₂ diagram. The theory was originally described by Rahn (15) and independently by Riley and Cournand (16) to depict the effect of ventilation-perfusion heterogeneity on inefficiency of O₂ and CO₂ exchange. The CO₂-O₂ diagram shows a curve describing the PO₂ of all possible alveoli in the lungs with a given inspired gas and mixed venous blood PO₂ and PCO₂. The alveolar points for the curve are calculated by assuming an identical gas exchange ratio for both alveolar gas and pulmonary capillary blood in each alveolus. A brief explanation of the method used for constructing such a curve can be found in Hlastala and Berger (17).

The hypothetical CO₂-O₂ diagram for a patient with the H pattern of ventilation-perfusion distribution is shown with a four-compartment lung (shunt, low ventilation-perfusion, high ventilation-perfusion, and dead space) in Figure 3. The alveolar point (A) is determined from a ventilation-weighted mean of the values from the two alveolar regions that receive ventilation (joined by the solid line). The mixed exhaled point (E) is the ventilation-weighted average of gas from the alveolar point and the dead space. The arterial point (a) is determined from the perfusion-weighted mean of blood gas contents coming from the three regions that receive perfusion (not obtainable graphically). The horizontal difference on the x-axis between (A) and (a) is the AaPO₂, in this case equal to 41.5 mm Hg. Notice the large vertical difference between (a) and (A) on the y-axis indicating a large aAPCO₂, which is a result of the substantial ventilation going to the high ventilation-perfusion region.

View larger version (15K): [in this window] [in a new window]   View larger version (16K): [in this window] [in a new window]   Figure 3.   Hypothetical O2-CO2 diagram of patients with COPD with the H (i.e., high A/) pattern of A/ heterogeneity in a four-compartment lung before ( A) and after (B) lung volume reduction surgery assuming the high A/ region was resected. Relative distributions of ventilation and perfusion to the four regions prior to LVRS ( A) are taken from the data of Wagner and colleagues (13). The location of the alveolar gas concentrations (point A) is determined from a ventilation-weighted mean of the values from the two alveolar regions that receive ventilation. The arterial gas concentrations (point a) is determined from the perfusion-weighted mean of blood gas contents coming from the three regions that receive perfusion. The mixed exhaled gas concentrations (point E) are determined from the perfusion-weighted average of gas from the alveolar point and the dead space. The horizontal difference on the x-axis represents the AaPO2. The vertical difference on the y-axis represents the aAPCO2. After resection the AaPO2 decreases, the PaO2 increases, and the aAPCO2 and PaCO2 decrease.

The hypothetical CO₂-O₂ diagram for a patient with the L pattern of ventilation-perfusion distribution is shown with a four-compartment lung (shunt, low ventilation-perfusion, high ventilation-perfusion, and dead space) in Figure 4. For this example, the AaPO₂ is equal to 64 mm Hg. The AaPO₂ is larger in this case because a large fraction (45%) of the blood flow goes through a region with very low oxygen content (only slightly greater than mixed venous blood oxygen content). In the L pattern, the aAPCO₂ is relatively small because of the absence of a high ventilation-perfusion region.

View larger version (16K): [in this window] [in a new window]   View larger version (16K): [in this window] [in a new window]   Figure 4.   Hypothetical O2-CO2 diagram of patients with COPD with the L (i.e., low A/) pattern of A/ heterogeneity in a four- compartment lung before (A) and after (B) lung volume reduction surgery, assuming the low A/ region was resected. Relative distributions of ventilation and perfusion to the four regions prior to LVRS ( A) are taken from the data of Wagner and colleagues (13). The arterial, alveolar, and mixed expired gas concentrations were determined as described in Figure 3. Prior to LVRS the AaPO2 is increased, the PaO2 is decreased, and the PaCO2 is increased compared with the H pattern shown in Figure 3, as a result of the effects of a greater amount of perfusion going to the low A/ region. The a APCO2 is lower because of the lack of a high A/ region. Following LVRS the AaPO2, PaCO2 and aAPCO2 are decreased, and the PaO2 is increased.

When hypoxemia results from blood perfusing areas of low ventilation-perfusion, improvements in Pa_O2 can result only from: (1) redistributing ventilation away from high ventilation-perfusion and/or dead space regions toward regions of low ventilation-perfusion; (2) redistributing perfusion away from the low ventilation-perfusion regions toward regions of higher ventilation-perfusion; and/or (3) improving the P_O2. LVRS can improve or worsen gas exchange and can narrow or widen the AaPO₂, depending on the relative degree of ventilation and perfusion going to the resected areas and the resulting ventilation-perfusion of the regions to which ventilation and/or perfusion redistribute.

If perfusion greatly exceeded ventilation in an area that was resected, this low ventilation-perfusion region would be eliminated and whatever perfusion was going to the area would redistribute elsewhere (assuming that cardiac output remained unchanged) (Figure 4). If the perfusion were redistributed to an area of high ventilation-perfusion, the ventilation-perfusion ratio in that region would normalize and the resulting Pa_O2 and AaPO₂ would improve. If the perfusion redistributed to an area of low ventilation-perfusion, the ventilation-perfusion ratio in that region would worsen, reducing the Pa_O2 and increasing the AaPO₂. The Pa_O2, the AaPO₂ and the Pa_CO2 would improve, worsen, or remain unchanged depending on whether the ventilation-perfusion ratio of the region that was now receiving more perfusion ended up being higher, lower, or the same as the ventilation-perfusion ratio of the region of lung that was resected.

If the areas resected were receiving any substantive degree of perfusion, the pulmonary arterial pressure might increase (as has been suggested by preliminary data reported by Weg and colleagues [18]).

If ventilation greatly exceeded perfusion and if resection removed this high ventilation-perfusion region, ventilation would redistribute to other areas (assuming that alveolar ventilation [A] remained unchanged). If A redistributed to an area of low ventilation-perfusion, the ventilation-perfusion ratio in this region would improve, with resulting improvements in the Pa_O2, AaPO₂, and Pa_CO2 (Figure 4). If A redistributed to areas of high ventilation-perfusion and/or dead space, the ventilation-perfusion in these region would increase further. As above, the Pa_O2, the AaPO₂, and the Pa_CO2 would improve, worsen, or remain unchanged, depending on whether the ventilation-perfusion ratio of the region that received more alveolar ventilation ended up being higher, lower, or the same as the ventilation-perfusion ratio of the region of lung that was resected.

If the areas resected received any substantive degree of alveolar ventilation, the resulting Pa_CO2 could increase if the mechanics of the remaining lung precluded a sufficient increase in its alveolar ventilation.

For the hypothetical lungs shown in Figures 3A and 4A, resection of the abnormal region, allowing all alveolar perfusion and ventilation to go to the remaining alveolar region, results in a much more homogeneous ventilation-perfusion distribution and a marked improvement in the AaPO₂. In the H pattern, lung resection of the high ventilation-perfusion regions results in a reduction of AaPO₂ from 41.5 to 12.2 mm Hg (Figure 3B). In the L pattern, lung resection of the low ventilation-perfusion regions results in a reduction of AaPO₂ from 64 to 3.4 mm Hg (Figure 4B). In the relatively homogeneous lungs shown in Figure 3B and Figure 4B, the AaPO₂ is comparably small, but is larger in Figure 3B because of the greater assumed shunt (3% versus 1%). If any perfusion were redistributed to this shunt region, gas exchange would worsen.

In both of the cases illustrated in Figures 3 and 4, resection of any portion of the lung deviating most from the normal ventilation-perfusion of approximately 1.0 narrows the remaining ventilation-perfusion heterogeneity. If the resected sections of the lung consisted of relatively normal ventilation- perfusion regions, and if the perfusion and/or ventilation of these regions were redistributed such that ventilation-perfusion heterogeneity of the remaining lung worsened, the AaPO₂ would increase and the Pa_O2 would fall.

Finally, many centers routinely obtain perfusion lung scans prior to LVRS in an attempt to target areas for resection that have little to no perfusion. Accordingly, the fraction of total perfusion that might be redirected following resection generally could be exceedingly small, and the resulting changes in ventilation-perfusion could be trivial.

Changes in P_O2 . Changes in P_O2 would affect gas exchange in areas of low ventilation-perfusion rates and/or shunt. Although P_O2 data have not been reported in patients undergoing LVRS, Albert and colleagues (19) found values that were below normal (35.8 ± 0.7 mm Hg, with venous oxygen saturation of 66.9 ± 1.0%) in 30 patients with stable COPD who had somewhat less severe airflow limitation than those reported here. P_O2 could be low in patients undergoing LVRS and might improve or worsen following the procedure, if the operation altered work of breathing (9, 11, 20, 21) and/or cardiac output and oxygen delivery (as a function of changes in mean pleural pressure and the level of auto-positive end- expiratory pressure (9).

In summary, LVRS can improve and worsen arterial blood gases, and the changes do not correlate with the effects of the operation on lung mechanics. Patients being evaluated for the procedure should be advised that, although oxygenation might improve following the operation, it is possible, and perhaps equally likely, that it will worsen. They could also be advised that, based on the information that is presently available, it is not possible to predict prior to the procedure who will have an improvement in oxygenation and who will not.