INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Keywords: lung volume reduction; outcome assessment; pulmonary emphysema; tomography (x-ray computed)

Computed tomography (CT) of the chest was first used to detect and quantify pulmonary emphysema by Hayhurst and colleagues, who analyzed the cumulative frequency distribution of the EMI numbers produced by the disease (1). Their report led to rapid expansion of both qualitative and quantitative assessments of emphysema with CT scans (2). In a subsequent group of studies, Müller and others based their estimates on a predetermined "density mask" measured in Hounsfield Units (HU), which was correlated with the pathologic findings in lungs (2). These CT measurements of lung density have also been used to calculate lung expansion, expressed as milliliters of gas per gram of lung tissue (ml/g) (8). This approach has the advantage of estimating emphysema in terms of its definition, which is the expansion of air spaces beyond normal, with destruction of their walls (14). Coxson and colleagues estimated that the normal lung at TLC contains a maximum of 6.0 ml gas/g lung tissue, and showed that lung expanded from 6.0 to 10.2 ml/g contained emphysematous lesions of less than 5 mm in diameter (mild/moderate emphysema), and that lung expanded beyond 10.2 ml/g contained larger lesions (severe emphysema) (8). Two recent CT studies have also shown that emphysema is more extensive in the inner (core) region than in the outer (rind) region of the lung, and that the extent of emphysema in the core was more closely associated with deterioration in lung function (11, 12). The present study describes the application of this technique in selecting patients for lung volume reduction surgery (LVRS).

LVRS is a palliative procedure for patients who have advanced emphysema, and a number of studies have shown symptomatic and functional improvement following LVRS (15). Unfortunately, not all patients have equivalent responses to LVRS (19, 22, 35), and investigators are divided with respect to preoperative anatomic and physiologic selection criteria for the procedure. We have recently shown that the volume of severe emphysema present in the preoperative CT scan correlates with the magnitude of functional improvement after LVRS (20). Other studies of the radiologic distribution of emphysema suggest that heterogeneously distributed disease, predominantly located in the upper lung zones, may predict a better response to surgery (26), but correlation of these indices with response is at best moderate.

This report describes a new index that quantitates the volume of emphysema with a central core versus that with a peripheral rind distribution, which we believe may be an indicator of disease that is surgically accessible and more amenable to an effective LVRS procedure. Such indices, if accurately indicative, can be used in the analysis of data already collected in multicenter trials, such as the National Emphysema Treatment Trial (NETT) (36).

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subjects

The analysis was performed on 21 patients (16 males and 5 females), who underwent bilateral LVRS either through a video-assisted approach (n = 17) or through median sternotomy (n = 4) done with previously reported techniques (15) between June 1994 and June 1997. Patients were selected if they could complete the previously described radiologic, physiologic, and cardiopulmonary exercise tests (CPX) at baseline (before surgery) and at 3-mo follow up (20, 21). Informed consent was obtained from each patient, and the study was approved by the institutional review board of the University of Pittsburgh Medical Center.

Quantitative CT Analysis

The subjects in the study received conventional, non-contrast-enhanced CT scans (10-mm-thick contiguous slices, with a 33.5 ± 2.4 cm field of view and standard reconstruction algorithm) on a GE 9800 Highlight Advantage CT scanner (General Electric Medical Systems, Milwaukee, WI) approximately 1 wk before surgery and 3 mo after surgery. Scans were performed with the patient in the supine position during breathholding at full inspiration.

The image data were transferred to a personal computer and analyzed with customized software. The program used for this is based on previously described methods (8, 11). Briefly, the lung parenchyma was segmented from both the chest wall and large central vessels through use of a contour-following algorithm and a CT threshold of 500 HU (8). The lung was then separated into core and rind regions, with the rind region defined as the peripheral 50% of the lung area and the remaining area defined as the core region (11). The boundary line between the core and rind regions had a constant distance from the lung surface (Figure 1) (11). Emphysema was defined as lung expanded beyond the normal range as estimated from measurements of lung density. Emphysematous lesions of more than 5-mm diameter are correlated with lung expanded beyond 10.2 ml/g (which corresponds to less than 910 HU), and lesions of less than 5-mm diameter are correlated with lung regions inflated between 6.0 and 10.2 ml/g (910 HU and 856 HU) (8). Total lung volume, the volume of mild/ moderate and severe emphysematous lesions (lung expanded beyond 10.2 ml/g and between 6.0 and 10.2 ml/g), and the volume of normal (lung expanded by less than 6.0 ml/g) were calculated.

View larger version (137K): [in this window] [in a new window]   Figure 1.   Division of the lung. Lung was divided into inner (core) and outer (rind) regions.

All CT slices cranial to the bifurcation of the trachea were considered to consist of the upper (cephalad) region of the lung, whereas slices below the tracheal carina were considered as representing the lower (caudal) region of the lung. This procedure divided the lungs into four regions (upper-core, upper-rind, lower-core, and lower-rind), and provided the volume of each region that was occupied by emphysematous lesions.

Statistical Analysis

All data are expressed as mean ± SD. Differences between groups were evaluated with a paired two-tailed t test. Univariate (linear) regression analysis and stepwise forward multiple regression analysis were used to evaluate the relationship between the CT parameters and the changes in measures from baseline to after surgery. A value of p < 0.05 was considered significant.

Additional information of the methodology in the study can be accessed in the article's online data supplement, accessible at www. atsjournals.org.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The age of the patients was 62.3 ± 7.7 (mean ± SD) yr. The patients' pulmonary function tests and maximal wattage measured by CPX before and after LVRS are shown in Table 1. These data show significant improvements in FEV₁, FVC, TLC, inspiratory capacity (IC), FRC, and RV, although there was no significant change in the diffusing capacity of carbon monoxide (DLCO). The maximal wattage during CPX increased significantly after LVRS.

Table 2 shows the regional distribution of emphysematous lesions and normal lung as assessed with pre- and postoperative CT, where the total lung volume as assessed with CT corresponds to 100%. Because the upper lung volume was smaller than the lower (2,086 ± 530 ml versus 5,128 ± 948 ml preoperatively, and 1,543 ± 386 ml versus 4,723 ± 932 ml postoperatively), the percentage of total lung volume for the upper region is smaller than that for the lower region. The fraction of lung expanded beyond 10.2 ml/g in the core was greater than in the rind both in the upper and lower regions (p < 0.0001). Conversely, the percentage of voxels containing normal lung (inflation < 6.0 ml/g), and that of lung containing mild/moderate lesions (inflation between 6.0 and 10.2 ml/g), was smaller in the core than in the rind (p < 0.001).

Table 3 shows the relationship between the change in wattage and the preoperative CT parameters. There was a positive correlation between the amount of upper lung expanded beyond 10.2 ml/g and the change in wattage following surgery. The amount of normal lung (% voxels showing inflation < 6.0 ml/g) and the amount of lung with mild/moderate disease (% voxels showing inflation between 6.0 and 10.2 ml/g) in the upper lung correlated negatively with improvement in exercise. The amount of total lung expanded beyond 10.2 ml/g correlated positively with the change of CPX as previously described (20). To evaluate the independent contribution of the different regions to these relationships, we used multiple regression analysis, entering variables in order of their r ² values when they were significant in the univariate regression analysis, while keeping severe emphysema (% voxels showing inflation > 10.2 ml/g) of total lung in the analysis. Stepwise multiple regression analysis showed that the addition of % voxels showing inflation > 10.2 ml/g in upper rind provided a significantly better fit (r ² = 0.462, p < 0.005) with the predicted change in wattage than did a linear equation using the total amount of emphysema as the sole independent variable (r² = 0.266).

The multiple regression analysis (Table 4) showed that the % voxels showing inflation > 10.2 ml/g in the upper rind also predicted the change in FEV₁ after surgery (r² = 0.356, p < 0.005), which was not significantly related to the % voxels showing inflation \> 10.2 ml/g in total lung.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

We have recently shown that the volume of severe emphysema present in the preoperative CT scan is a good predictor of outcome after LVRS (20). The present study adds value to this prediction by showing that LVRS has a better outcome when the emphysema is concentrated in the upper rind of the lung. We chose a change in wattage in CPX as the outcome measure after LVRS because it represents an integration of the function of the entire cardiopulmonary system (22). Several studies have shown that patients with a heterogeneous pattern of emphysema respond better to LVRS than do those with a homogeneous distribution (26). Our results confirm these previous reports and extend them by showing that patients who have more emphysema in the surgically accessible upper-rind region of the lung show the greatest benefit from LVRS.

In addition to the prediction provided by the improvement in wattage in CPX, the extent of severe emphysema (% voxels showing inflation > 10.2 ml/g) in the upper-rind of the lung also predicts the improvement in FEV₁ after LVRS (r ² = 0.356). This is consistent with the report of Maki and colleagues, who used preoperative chest roentgenograms to show that the semiquantitative score of upper-lung-zone predominance of emphysema correlated with the postoperative increase in FEV₁ (r ² = 0.46) (29). Slone and colleagues also used a semiquantitative CT score to show that upper lobe disease correlated with the change in FEV₁ (r² = 0.12) (27). An advantage of the present study is that the distribution of core versus rind disease is determined by computer.

This study was performed retrospectively, using conventional CT scanning techniques. Although many research and clinical institutions are now assessing emphysema with high-resolution CT (HRCT) because of improved edge detection, the high spatial frequency reconstruction algorithm used for HRCT decreases the signal-to-noise ratio. It has been shown that the density resolution of the CT scan is better with a thicker slice and a lower spatial frequency reconstruction (37), indicating that conventional CT scanning gives better estimates of lung density. The new multidetector row scanners allow scanning of both thick and thin images without increasing the radiation dose received by the patient. These instruments also have the ability to scan the entire lung during a single breathhold (38), which will further enhance the quantitative assessment of emphysema and provide a valuable means for documenting the natural history of the disease and measuring any response to treatment.

The preoperative CT data show that severe emphysematous lesions (% voxels showing inflation > 10.2 ml/g) were located predominantly in the core of the lung, which is in accordance with previous reports (11, 12). Conversely, mild/moderate emphysema (% voxels showing inflation between 6.0 and 10.2 ml/g) and normal lung tissue (% voxels showing inflation by < 6.0 ml/g) were more frequent in the rind of the lung. Although all 21 patients in the study had more extensive severe emphysema (% voxels showing inflation by > 10.2 ml/g) in the core of the lung, those who had relatively more emphysema in the rind had the greatest benefit from surgery.

Although there was a difference in the regional distribution of emphysema between the core and the rind of the lung after LVRS, it is difficult to compare the pre- and postoperative distribution of emphysema, because of the spatial rearrangement that occurs within the thorax after LVRS. Overall, we confirmed our previous study's finding that the decrease in total lung volume after surgery (7,214 ± 1,299 ml versus 6,266 ± 1,138 ml, p < 0.0001) was predominantly due to the decrease in severe emphysema (3,389 ± 1,203 ml versus 2,376 ± 947 ml, p < 0.0001) (20). Also, the volume of mild/moderate disease remained unchanged (1,955 ± 450 ml versus 1,914 ± 462 ml, p = 0.52), with a trend toward an increase in normal lung volume (1,870 ± 360 ml versus 1,976 ± 423 ml, p = 0.09), suggesting recruitment of relatively compressed lung after surgery (20).

Because the rind of the lung is more accessible to surgery than is the core, we assumed that this region is preferentially removed by LVRS, and that when it contains severe emphysema, there is a better outcome. In summary, the measurement of emphysema in the core and rind of the lung may be a useful indicator of disease that is surgically accessible and therefore more amenable to LVRS. This hypothesis should be confirmed by performing the analysis described here on data collected in large multicenter trials, such as the NETT (36).