点击显示 收起
1 From the Departments of Radiology (A.R.H., P.C.) and Pulmonary and Critical Care Medicine (E.P.I., J.J.R.), Brigham and Women’s Hospital, 75 Francis St, Boston, MA 02115. From the 2000 RSNA scientific assembly. Received February 14, 2001; revision requested March 20; revision received June 8; accepted July 5.
|
|
ABSTRACT |
|---|
|
Top |
|---|
PURPOSE: To compare the relationship between computer-derived and visually assessed ventilation-perfusion (V/Q) scintigraphy and computed tomographic (CT) scores in evaluating disease severity and distribution in identifying optimal candidates for lung volume reduction surgery (LVRS) and to correlate these radiologic indices with physiologic measures of outcome.
MATERIALS AND METHODS: In 39 patients, preoperative V/Q and CT scans were visually scored by two radiologists for disease severity and distribution. Results were compared with computer-derived scores for the same cohort. These indices were correlated with clinical improvement measured with forced expiratory volume in 1 second (FEV1), forced vital capacity (FVC), and ratio of FEV1 to FVC.
RESULTS: The disease distribution scores measured with the different methods correlated closely: computer-based and visually assessed CT scores (r = 0.89, P < .001), computer-based and visually assessed V/Q scores (r = 0.83, P < .001), visually assessed CT and V/Q scores (r = -0.50, P < .001), and computer-derived CT and V/Q scores (r = -0.57, P = .015). Similarly, a statistically significant correlation was noted between each of the radiologic methods and clinical outcome measurements (P < .001).
CONCLUSION: CT and V/Q preoperative assessment, with either visual scoring or computer-based algorithms, are nearly equivalent in their utility in predicting improvement in FEV1 measures.
Index terms: Emphysema, pulmonary, 60.751 • Lung, CT, 60.12115 • Lung, function, 60.919 • Lung, radionuclide studies, 60.12171 • Lung, surgery, 60.45
| INTRODUCTION |
|---|
| Top ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES |
|---|
LVRS is performed in an attempt to normalize the relationship between lung volume and the size of the chest wall, which ostensibly improves respiratory mechanics, the work of breathing, and ventilation-perfusion (V/Q) matching in the remaining lung (1,2). This is achieved by performing several nonsegmental wedge resections with the aim of reducing overall lung volume by 20%–30% (3).
Authors of prior reports (3,4) have indicated that postoperative improvement in lung function is related to improved elastic recoil. This has certainly been shown to be true following focal bullectomies in which normal lung tissue is compressed (5–8) and has been proposed as the primary mechanism for improvement in LVRS.
Currently, most patients evaluated for LVRS undergo extensive preoperative radiologic imaging, including chest radiography, chest computed tomography (CT), and V/Q scintigraphy (9,10). Together, these studies are intended to provide information as to the extent, severity, and distribution of disease. This information is then used to identify potential candidates for surgery and to target the surgical procedure so that the most diseased regions of lung are resected. Additionally, radiologic information combined with physiologic studies may serve as predictors of positive or negative outcomes (11–13).
The purpose of our study was twofold. First, we sought to compare the utility of computer-derived (quantitative) and visually assessed (qualitative) V/Q scintigraphic and CT indices of disease distribution and severity in identifying optimal candidates for LVRS. Second, we attempted to correlate radiologic indices with physiologic outcome variables in a way that may shed light on the specific mechanisms through which lung function improves following LVRS.
| MATERIALS AND METHODS |
|---|
| Top ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES |
|---|
Radiologic Evaluation and Scoring
CT imaging and scoring.—Thin-section CT scanning (for visual scoring) and helical CT scanning (for computer-based scoring) were performed in all 39 subjects with a Somatom Plus 4 scanner (Siemens, Iselin, NJ), with 1-mm beam collimation at 20-mm intervals during full inspiration. Helical images were acquired with 4-mm collimation. Images were reviewed at a window width of 1,500 HU and a level of -600 HU, appropriate for viewing the lung parenchyma.
The thin-section CT images were analyzed retrospectively and individually and were visually scored by two radiologists (A.R.H., P.C.) who were blinded to clinical and V/Q data. The visual scoring system described by Goddard et al (15) was used to obtain a qualitative CT score. Emphysematous destruction was identified as areas of low attenuation (ie, hypovascular regions) in the lungs. The lung was divided into two zones, an upper and a lower zone, for scoring on each side. The upper zone extended from the carina superiorly to the apex, and the lower zone extended from the carina to the lung bases. This approach was taken to enable the readers to explicitly ascribe disease to either upper or lower lobes. We used a four-point scale described by Goddard et al (15) to grade disease severity. A score of 1 represents destruction of 1%–25% of the lung by emphysema; a score of 2, destruction of 26%–50% of the lung; a score of 3, destruction of 51%–75% of the lung; and a score of 4, destruction of 76%–100% of the lung. The final score was achieved through agreement of the two readers or was given as an average if the readers differed by one point. A maximum score, representing severe disease in all zones (indicating marked diffuse emphysema) would be 16. Total scores were obtained by means of summing the scores from all zones. An upper lobe to lower lobe disease distribution score was calculated as the ratio of upper lobe to lower lobe scores.
In 17 of the 39 patients, computer-based quantitative assessment of disease severity and distribution with CT was available and was obtained by using a semiautomated attenuation-mask program (16). Attenuation mapping was performed on the entire volume of lung with the helical CT data set. Regions of interest encompassing both lungs were identified; care was taken to exclude mediastinal structures, major pulmonary vessels, and the chest wall. The area of each section occupied by voxels of less than -910 HU was calculated for each level (16–18). The area of low-attenuation voxels (ie, voxels of -910 HU or less) was divided by the total area of the lung to obtain the percentage of emphysematous lung (18). The threshold value of -910 HU for defining emphysematous lung has been previously described and validated (16). The percentage of areas of emphysematous destruction was derived for the whole lung and for upper and lower lung zones.
V/Q imaging and scoring.—The cohort also underwent preoperative V/Q scintigraphy with a large field of view scintillation camera (CAMSTAR; GE Medical Systems, Milwaukee, Wis), with a low-energy medium-resolution parallel-hole collimator. Each patient received 111 MBq (3 mCi) of technetium-labeled macroaggregated albumin (Brigham and Women’s Hospital, Boston, Mass). Approximately 400,000–500,000 particles were injected. Posterior, anterior, and left and right posterior oblique and lateral views were acquired for 600,000 counts. Counts were apportioned to the upper, middle, and lower lung zones by means of dividing the right and left lungs into three equally sized regions and measuring total counts in each of these regions. Lobar predominance was assessed by means of an index defined by the average number of counts in the upper lobes divided by the average number of counts in the middle and lower lobes. Thus a quantitative computer-based score was generated.
Additionally, the perfusion images were reviewed and individually and retrospectively scored by two radiologists (one board certified in nuclear medicine [P.C.] and the other fellowship trained in thoracic radiology [A.R.H.]) to obtain a qualitative visual score. The final score was obtained by means of a joint review of the two readers, or was given as an average of readings from the two if they differed by one point. The radiologists were blinded to the clinical and CT data. The scoring system used for visual assessment was described by Wang et al (11) and is as follows: The images were graded for heterogeneity, percentage of lung maximally perfused, and for upper lobe– versus lower lobe–predominant disease. Heterogeneity was defined as the regional variation in the severity of emphysema. Scores ranged from 1 to 10, with high scores representing target areas of severe emphysema and low scores representing diffuse disease or little variation in the severity of emphysema. The percentage of maximal perfusion was assessed as the amount of lung perfused at maximal counts. Lobar predominance was assessed on a scale of one to five, with a low score indicating predominantly upper lobe emphysema and a high score indicating predominantly lower lobe emphysema.
Physiologic Testing
Spirometery was performed by two pulmonologists (E.P.I., J.J.R.) both before and after LVRS with a rolling seal system spirometer (Mark V; Morgan, Haverhill, Mass). FEV1, FVC, and FEV1/FVC were measured within 1 hour of bronchodilator therapy and were recorded according to American Thoracic Society guidelines (19). Percent predicted values were derived from the population data of Knudson et al (20).
Data Analysis
Visually assessed and computer-derived scores of disease severity and distribution for both V/Q and CT imaging were compared with one another and with the patients’ physiologic responses 6 months after surgery as measured with FEV1, FVC, and FEV1/FVC values. Results are reported as the mean ± SD, and comparisons between preoperative and postoperative values were performed with a paired t test analysis. Univariate and multivariate regression analyses were performed with a least squares approach (Statmost Software, Stamford, Conn). Statistical significance was defined as a P value less than .05.
| RESULTS |
|---|
| Top ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES |
|---|
|
There was a close correlation between disease distribution scores as measured with the different methodologies. summarizes the relationships between quantitative and qualitative V/Q scores, quantitative and qualitative CT scores, quantitative CT and quantitative V/Q scores, and qualitative CT and qualitative V/Q scores. In each instance, statistically significant associations between the distinct scoring systems were observed. Specifically, comparison of computer-based and visually assessed scores yielded the following correlations: computer-based and visually assessed CT scores (r = 0.89, P < .001), computer-based and visually assessed V/Q scores (r = 0.83, P < .001), visually assessed CT and V/Q scores (r = -0.50, P < .001), and computer-based CT and V/Q scores (r = -0.57, P = .015). Representative V/Q and CT images from a patient in our study demonstrate that both imaging modalities show similar patterns of disease distribution. It should be noted that all relationships involving quantitative CT data included only 17 patients. For this reason, correlations between quantitative and qualitative V/Q scores and between qualitative V/Q and qualitative CT scores were evaluated for both the entire cohort (n = 39) and for the subgroup (n = 17) in whom quantitative CT data was available. Relationships observed between the evaluations of the subgroup of 17 patients were statistically similar to those observed in the larger cohort.
|
|
|
|
|
|
|
|
|
|
|
| DISCUSSION |
|---|
| Top ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES |
|---|
Although many studies have focused on developing reliable criteria for applying either CT or radionuclide scintigraphic imaging to the identification of patients with upper lobe disease, few studies have compared the two approaches for screening patients in the same emphysema population. Such a comparison would help determine whether CT and V/Q scanning provide distinct and complementary or largely overlapping information and whether one approach is truly superior in identifying candidates for LVRS, potentially eliminating the need for both studies. The present study directly compares the utility of CT and V/Q scanning in the same LVRS cohort, examines how computer algorithm–based quantitative scoring compares with established visual scoring methods, and examines the relative utility of each modality and scoring method in predicting successful outcomes of LVRS.
Comparison of visually assessed CT and V/Q scores of disease distribution demonstrated a significant correlation between the two modalities (n = 39, r = 0.50, P < .001) that was similar to observations made by other investigators (21). Furthermore, computer-based (quantitative) scores of disease distribution at CT correlated closely with visually assessed (qualitative) scores (n = 17, r = 0.89, P < .001). Computer-derived and visually assessed scores of disease distribution on V/Q scans were also closely correlated (n = 39, r = 0.83, P < .001). These results are important for two reasons. First, they suggest that some redundancy does exist between the information obtained at CT and that obtained at V/Q, although the two modalities clearly do not provide identical information. Second, in a finding that is perhaps of greater clinical importance, the results suggest that assessments of disease distribution and severity by experienced readers on the basis of CT and V/Q images are similar to assessments generated with computer-based algorithms.
Visually assessed disease distribution scores generated on the basis of CT data and those generated on the basis of V/Q data were nearly equivalent with respect to predicting improvement in FEV1 at 6-month postoperative follow-up. Computer-based quantitative algorithms for both CT and V/Q scans had similar predictive utility, although the correlations between the change in FEV1 and quantitative CT scores did not achieve statistical significance because the latter were available for only a fraction of the study cohort. Although only 16%–25% of the change in FEV1 could be predicted with preoperative radiologic indices, these values are similar to those previously reported by other investigators (12,21). These weak correlations confirm previous observations that suggest that although upper lobe–predominant disease is an important determinant of physiologic responsiveness to LVRS, many patients with other radiologic patterns of disease improve in response to surgery. Our results also indicate that the lack of close correlation between CT and V/Q indices does not result in one modality being clearly superior to the other in terms of its utility in predicting LVRS outcome. Thus, the differences in functional and physiologic information provided by these two approaches do not contribute information that substantially impacts on their ability to predict favorable responses to LVRS.
Our observations also have interesting physiologic implications with respect to how LVRS actually works to improve lung function. The presence of upper lobe–predominant disease identified with any of these approaches correlated not only with physiologic improvements in the lung, but also with a particular pattern of improvement. Specifically, FEV1, the most common physiologic index of improvement used to evaluate LVRS outcomes, is determined by two distinct factors: (a) the ratio of FEV1 to the FVC, which is an index of obstruction, and (b) the absolute FVC, which is an index of total functional volume within the chest cavity. If radiologic indices helped predict a favorable outcome to LVRS through a mechanism that involved a decrease in resistance to airflow, these indices would be expected to correlate with changes in the FEV1/FVC ratio. Conversely, if radiologic indices helped identify patients in whom function improved largely through an increase in functional lung volume, the indices would correlate with an increase in FVC. Upper lobe disease predominance measured by any approach correlated significantly with increases in FVC but not with increases in the FEV1/FVC ratio. These findings suggest that patients with upper lobe disease tend to experience improvement in their respiratory function primarily as a result of an increase in functional lung volume as reflected in an increase in vital capacity. Our observations are consistent with the mechanistic arguments proposed by Fessler and Permutt (22), which suggest that surgical resizing of the hyperinflated lung increases total lung capacity relative to residual volume and improves maximal expiratory flow through increasing recoil pressures, without changing obstruction to airflow.
The findings reported here are largely consistent with those of other investigators, although several important differences are worth noting. Correlations between CT scan indices and improvement in FEV1 at 6-month follow-up were not as strong in our study as those observed by Slone et al (12). This apparent discrepancy may be due to differences in patient selection, incidence of postoperative complications, or other potential biases. Our results are more consistent with the observations of Wang et al (11) and Thurnheer et al (21), who demonstrated less pronounced but still statistically significant relationships between preoperative radiologic indices of disease distribution and postoperative improvement in lung function. However, in contrast to our study, the results of Thurnheer et al showed that only CT indices, not V/Q indices, correlated significantly with postoperative improvement in lung function. Nevertheless, the same relationship between preoperative imaging indices and change in FVC was observed, supporting the argument that upper lobe-predominant disease represents a physiologic state in which tissue resection improves function by allowing "recruitment" of vital capacity from remaining regions of lung through lung/chest wall "resizing."
We recognize the limitations of our study, namely its small sample size and retrospective nature. However, our data confirm previous observations that patients with upper lobe–predominant disease on preoperative radiologic studies are most likely to respond to LVRS. Furthermore, our data suggest that information obtained either with CT alone or with V/Q scintigraphy alone is sufficient for screening patients for LVRS, even though the two studies do not provide identical information. CT does yield additional information, however, such as the presence of unsuspected malignancy, bronchiectasis, pleural disease, or pulmonary fibrosis, which may dramatically affect whether or not LVRS is performed.
Our results demonstrate that because visually assessed and computer-derived scores for emphysema at CT and V/Q scanning correlate, there is no need to perform the more costly and time-consuming computer analysis. Also, our data indicated that after LVRS, lung function in patients with upper lobe disease improves as a result of the removal of tissue that was dysfunctional and not contributing to vital capacity. Resection of these areas increases the volume of functional tissue in the remaining lung tissues within the chest, which is specifically reflected in an increase in vital capacity measured at spirometry.
| REFERENCES |
|---|
| Top ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES |
|---|